From ccabd1644316387b5d059ee00b37a9386c0e5021 Mon Sep 17 00:00:00 2001 From: Yani Guan Date: Thu, 21 Nov 2024 15:11:22 -0800 Subject: [PATCH] delete old fashion reviews.bib and others.bib --- source/_data/others.bib | 7565 -------------------------------------- source/_data/reviews.bib | 715 ---- source/papers/others.md | 8 - source/papers/reviews.md | 7 - 4 files changed, 8295 deletions(-) delete mode 100644 source/_data/others.bib delete mode 100644 source/_data/reviews.bib delete mode 100644 source/papers/others.md delete mode 100644 source/papers/reviews.md diff --git a/source/_data/others.bib b/source/_data/others.bib deleted file mode 100644 index f5540a12..00000000 --- a/source/_data/others.bib +++ /dev/null @@ -1,7565 +0,0 @@ - - -405 Other papers with abstract. - -@article{acharEfficientlyTrainedDeep2021a, - abstract = {We have developed an accurate and efficient deep-learning potential (DP) for graphane, which is a fully hydrogenated version of graphene, using a very small training set consisting of 1000 snapshots from a 0.5 ps density functional theory (DFT) molecular dynamics simulation at 1000 K. We have assessed the ability of the DP to extrapolate to system sizes, temperatures, and lattice strains not included in the training set. The DP performs surprisingly well, outperforming an empirical many-body potential when compared with DFT data for the phonon density of states, thermodynamic properties, velocity autocorrelation function, and stress−strain curve up to the yield point. This indicates that our DP can reliably extrapolate beyond the limit of the training data. We have computed the thermal fluctuations as a function of system size for graphane. We found that graphane has larger thermal fluctuations compared with graphene, but having about the same out-of-plane stiffness.}, - author = {Achar, Siddarth K. and Zhang, Linfeng and Johnson, J. Karl}, - date = {2021-07-15}, - doi = {10/gmfwwb}, - issn = {1932-7447, 1932-7455}, - journaltitle = {The Journal of Physical Chemistry C}, - langid = {english}, - number = {27}, - pages = {14874--14882}, - shortjournal = {J. Phys. Chem. C}, - title = {Efficiently {{Trained Deep Learning Potential}} for {{Graphane}}}, - url = {https://pubs.acs.org/doi/10.1021/acs.jpcc.1c01411}, - urldate = {2021-08-10}, - volume = {125} -} - -@inproceedings{andersonCormorantCovariantMolecular2019, - abstract = {We propose Cormorant, a rotationally covariant neural network architecture for learning the behavior and properties of complex many-body physical systems. We apply these networks to molecular systems with two goals: learning atomic potential energy surfaces for use in Molecular Dynamics simulations, and learning ground state properties of molecules calculated by Density Functional Theory. Some of the key features of our network are that (a) each neuron explicitly corresponds to a subset of atoms; (b) the activation of each neuron is covariant to rotations, ensuring that overall the network is fully rotationally invariant. Furthermore, the non-linearity in our network is based upon tensor products and the Clebsch-Gordan decomposition, allowing the network to operate entirely in Fourier space. Cormorant significantly outperforms competing algorithms in learning molecular Potential Energy Surfaces from conformational geometries in the MD-17 dataset, and is competitive with other methods at learning geometric, energetic, electronic, and thermodynamic properties of molecules on the GDB-9 dataset.}, - annotation = {WOS:000535866906022}, - author = {Anderson, Brandon and Hy, Truong-Son and Kondor, Risi}, - booktitle = {Advances in {{Neural Information Processing Systems}} 32 (Nips 2019)}, - date = {2019}, - editor = {Wallach, H. and Larochelle, H. and Beygelzimer, A. and d'Alche- Buc, F. and Fox, E. and Garnett, R.}, - issn = {1049-5258}, - langid = {english}, - location = {{La Jolla}}, - options = {useprefix=true}, - publisher = {{Neural Information Processing Systems (nips)}}, - shorttitle = {Cormorant}, - title = {Cormorant: {{Covariant Molecular Neural Networks}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {32} -} - -@article{Andolina2020, - abstract = {We show that a deep-learning neural network potential (DP) based on density functional theory (DFT) calculations can well describe Cu-Zr materials, an example of a binary alloy system, that can coexist in as ordered intermetallic and as an amorphous phase. The complex phase diagram for Cu-Zr makes it a challenging system for traditional atomistic force-fields that cannot accurately describe the different properties and phases. Instead, we show that a DP approach using a large database with ∼300k configurations can render results generally on par with DFT. The training set includes configurations of pristine and bulk elementary metals and intermetallic structures in the liquid and solid phases in addition to slab and amorphous configurations. The DP model was validated by comparing bulk properties such as lattice constants, elastic constants, bulk moduli, phonon spectra, and surface energies to DFT values for identical structures. Furthermore, we contrast the DP results with values obtained using well-established two embedded atom method potentials. Overall, our DP potential provides near DFT accuracy for the different Cu-Zr phases but with a fraction of its computational cost, thus enabling accurate computations of realistic atomistic models, especially for the amorphous phase.}, - author = {Andolina, Christopher M. and Williamson, Philip and Saidi, Wissam A.}, - date = {2020-04}, - doi = {10.1063/5.0005347}, - journaltitle = {Journal of Chemical Physics}, - number = {15}, - publisher = {{American Institute of Physics Inc.}}, - title = {Optimization and Validation of a Deep Learning {{CuZr}} Atomistic Potential: {{Robust}} Applications for Crystalline and Amorphous Phases with near-{{DFT}} Accuracy}, - volume = {152} -} - -@article{andolinaRobustMultiLengthScaleMachine2021, - abstract = {Materials composed of Ag, Au, and Ag–Au alloys remain of great interest despite decades of intense research scrutiny. We interpret these efforts as an impetus for developing robust, accurate, and relatively fast computational methods for modeling these materials. Herein, we describe the training, development, and validation of a machine learning deep neural-network potential (DNP) for improved modeling of Ag–Au systems. This DNP was iteratively trained using density functional theory (DFT) to produce a robust multi-length-scale potential, which yields results comparable to DFT on a wide range of properties such as equilibrium and nonequilibrium lattices, mechanical properties, and defect energies. Further, this DNP can well describe adatom (Ag or Au) energy barriers for diffusion on \{100\}-, \{110\}-, and \{111\}-terminated surfaces (Ag or Au), in agreement with previously reported works. We utilized the DNP to study the nucleation and growth of simulated seeded core–shell Ag and Au nanoparticles (NP). We show that both nanoalloys grow such that \{111\} facets significantly increase at the expense of the \{100\} ones. In contrast, the Ag core NP is found to have a more disordered inner structure than the Au one and that Ag adatoms in Au@Ag NP have a more pronounced penetration power than Au in Ag@Au NP. These findings are rationalized in terms of adatom adsorption and diffusion energies.}, - author = {Andolina, Christopher M. and Bon, Marta and Passerone, Daniele and Saidi, Wissam A.}, - date = {2021-07-29}, - doi = {10/gmdj4k}, - issn = {1932-7447}, - journaltitle = {The Journal of Physical Chemistry C}, - publisher = {{American Chemical Society}}, - shortjournal = {J. Phys. Chem. C}, - title = {Robust, {{Multi}}-{{Length}}-{{Scale}}, {{Machine Learning Potential}} for {{Ag}}–{{Au Bimetallic Alloys}} from {{Clusters}} to {{Bulk Materials}}}, - url = {https://doi.org/10.1021/acs.jpcc.1c04403}, - urldate = {2021-08-11} -} - -@article{andradeFreeEnergyProton2020, - abstract = {TiO2 is a widely used photocatalyst in science and technology and its interface with water is important in fields ranging from geochemistry to biomedicine. Yet, it is still unclear whether water adsorbs in molecular or dissociated form on TiO2 even for the case of well-defined crystalline surfaces. To address this issue, we simulated the TiO2-water interface using molecular dynamics with an ab initio-based deep neural network potential. Our simulations show a dynamical equilibrium of molecular and dissociative adsorption of water on TiO2. Water dissociates through a solvent-assisted concerted proton transfer to form a pair of short-lived hydroxyl groups on the TiO2 surface. Molecular adsorption of water is Delta F = 8.0 +/- 0.9 kJ mol(-1) lower in free energy than the dissociative adsorption, giving rise to a 5.6 +/- 0.5\% equilibrium water dissociation fraction at room temperature. Due to the relevance of surface hydroxyl groups to the surface chemistry of TiO2, our model might be key to understanding phenomena ranging from surface functionalization to photocatalytic mechanisms.}, - annotation = {WOS:000519240000031}, - author = {Andrade, Marcos F. Calegari and Ko, Hsin-Yu and Zhang, Linfeng and Car, Roberto and Selloni, Annabella}, - date = {2020-03-07}, - doi = {10.1039/c9sc05116c}, - issn = {2041-6520}, - journaltitle = {Chemical Science}, - langid = {english}, - location = {{Cambridge}}, - number = {9}, - pages = {2335--2341}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Chem. Sci.}, - title = {Free Energy of Proton Transfer at the Water-{{TiO2}} Interface from Ab Initio Deep Potential Molecular Dynamics}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{andreaniHydrogenDynamicsSupercritical2020, - abstract = {In this work, an investigation of supercritical water is presented combining inelastic and deep inelastic neutron scattering experiments and molecular dynamics simulations based on a machine-learned potential of ab initio quality. The local hydrogen dynamics is investigated at 250 bar and in the temperature range of 553-823 K, covering the evolution from subcritical liquid to supercritical gas-like water. The evolution of libration, bending, and stretching motions in the vibrational density of states is studied, analyzing the spectral features by a mode decomposition. Moreover, the hydrogen nuclear momentum distribution is measured, and its anisotropy is probed experimentally. It is shown that hydrogen bonds survive up to the higher temperatures investigated, and we discuss our results in the framework of the coupling between intramolecular modes and intermolecular librations. Results show that the local potential affecting hydrogen becomes less anisotropic within the molecular plane in the supercritical phase, and we attribute this result to the presence of more distorted hydrogen bonds.}, - annotation = {WOS:000589920000077}, - author = {Andreani, Carla and Romanelli, Giovanni and Parmentier, Alexandra and Senesi, Roberto and Kolesnikov, Alexander and Ko, Hsin-Yu and Andrade, Marcos F. Calegari and Car, Roberto}, - date = {2020-11-05}, - doi = {10.1021/acs.jpclett.0c02547}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {21}, - pages = {9461--9467}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - title = {Hydrogen {{Dynamics}} in {{Supercritical Water Probed}} by {{Neutron Scattering}} and {{Computer Simulations}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{angActiveLearningAccelerates2020, - abstract = {Modeling dynamical e -ects in chemical reactions, such as post-transition state bi-furcation, requires ab initio molecular dynamics simulations due to the breakdown of simpler static models like transition state theory. However, these simulations tend to be restricted to lower-accuracy electronic structure methods and scarce sampling be- cause of their high computational cost. Here, we report the use of statistical learning to accelerate reactive molecular dynamics simulations by combining high-throughput ab initio calculations, graph-convolution interatomic potentials and active learning. This pipeline was demonstrated on an ambimodal trispericyclic reaction involving 8,8-dicyanoheptafulvene and 6,6-dimethylfulvene. With a dataset size of approximately 31,000 M062X/def2-SVP quantum mechanical calculations, the computational cost of exploring the reactive potential energy surface was reduced by an order of magnitude. Thousands of virtually costless picosecond-long reactive trajectories suggest that post-transition state bifurcation plays a minor role for the reaction in vacuum. Furthermore, a transfer-learning strategy effectively upgraded the potential energy surface to higher levels of theory ((SMD-)M06-2X/def2-TZVPD in vacuum and three other solvents, as well as the more accurate DLPNO-DSD-PBEP86 D3BJ/def2-TZVPD) using about 10\% additional calculations for each surface. Since the larger basis set and the dynamic correlation capture intramolecular non-covalent interactions more accurately, they uncover longer lifetimes for the charge-separated intermediate on the more accurate potential energy surfaces. The character of the intermediate switches from entropic to thermodynamic upon including implicit solvation effects, with lifetimes increasing with solvent polarity. Analysis of 2,000 reactive trajectories on the chloroform PES shows a qualitative agreement with the experimentally-reported periselectivity for this reaction. This overall approach is broadly applicable and opens a door to the study of dynamical effects in larger, previously-intractable reactive systems.}, - author = {Ang, Shi Jun and Wang, Wujie and Schwalbe-Koda, Daniel and Axelrod, Simon and Gomez-Bombarelli, Rafael}, - date = {2020}, - title = {Active Learning Accelerates Ab Initio Molecular Dynamics on Pericyclic Reactive Energy Surfaces}, - url = {https://chemrxiv.org/engage/api-gateway/chemrxiv/assets/orp/resource/item/60c74a88bdbb893c46a393be/original/active-learning-accelerates-ab-initio-molecular-dynamics-on-pericyclic-reactive-energy-surfaces.pdf} -} - -@article{angActiveLearningAccelerates2021, - abstract = {Modeling dynamical effects in chemical reactions typically requires ab initio molecular dynamics (AIMD) simulations due to the breakdown of transition state theory (TST). Reactive AIMD simulations are limited to lower-accuracy electronic structure methods and weak statistics because quantum mechanical energies and forces must be evaluated at femtosecond time resolution over many replicas. We report a data-driven pipeline that allows for the treatment of dynamical effects with the same level of theory and overall cost as that of TST approaches. High-throughput ab initio calculations and autonomous data acquisition are coupled to graph convolutional neural-network interatomic potentials, allowing for inexpensive reactive AIMD simulations at quantum mechanical accuracy. We demonstrate the approach by accurately simulating post-TS dynamical effects in three distinct pericyclic reactions, including a challenging trispericyclic reaction with a complex bifurcating potential energy surface. This approach is broadly applicable to understanding dynamical effects and predicting reaction outcomes in large, previously intractable systems.}, - author = {Ang, Shi Jun and Wang, Wujie and Schwalbe-Koda, Daniel and Axelrod, Simon and Gómez-Bombarelli, Rafael}, - date = {2021-03}, - doi = {10/gmgdj2}, - issn = {24519294}, - journaltitle = {Chem}, - langid = {english}, - number = {3}, - pages = {738--751}, - shortjournal = {Chem}, - title = {Active Learning Accelerates Ab Initio Molecular Dynamics on Reactive Energy Surfaces}, - url = {https://linkinghub.elsevier.com/retrieve/pii/S2451929420306410}, - urldate = {2021-08-11}, - volume = {7} -} - -@article{atlasEmbeddingQuantumStatistical2021a, - abstract = {Quantum-mechanically driven charge polarization and charge transfer are ubiquitous in biomolecular systems, controlling reaction rates, allosteric interactions, ligand-protein binding, membrane transport, and dynamically driven structural transformations. Molecular dynamics (MD) simulations of these processes require quantum mechanical (QM) information in order to accurately describe their reactive dynamics. However, current techniques-empirical force fields, subsystem approaches, ab initio MD, and machine learning-vary in their ability to achieve a consistent chemical description across multiple atom types, and at scale. Here we present a physics-based, atomistic force field, the ensemble DFT charge-transfer embedded-atom method, in which QM forces are described at a uniform level of theory across all atoms, avoiding the need for explicit solution of the Schrodinger equation or large, precomputed training data sets. Coupling between the electronic and atomistic length scales is effected through an ensemble density functional theory formulation of the embedded-atom method originally developed for elemental materials. Charge transfer is expressed in terms of ensembles of ionic state basis densities of individual atoms, and charge polarization, in terms of atomic excited-state basis densities. This provides a highly compact yet general representation of the force field, encompassing both local and system-wide effects. Charge rearrangement is realized through the evolution of ensemble weights, adjusted at each dynamical time step via chemical potential equalization.}, - annotation = {WOS:000648873600024}, - author = {Atlas, Susan R.}, - date = {2021-05-06}, - doi = {10.1021/acs.jpca.1c00164}, - issn = {1089-5639}, - journaltitle = {Journal of Physical Chemistry A}, - langid = {english}, - location = {{Washington}}, - number = {17}, - pages = {3760--3775}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. A}, - title = {Embedding {{Quantum Statistical Excitations}} in a {{Classical Force Field}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {125} -} - -@article{balyakinDeepMachineLearning2020, - abstract = {The use of machine learning to develop neural network potentials (NNP) representing the interatomic potential energy surface allows us to achieve an optimal balance between accuracy and efficiency in computer simulation of materials. A key point in developing such potentials is the preparation of a training dataset of ab initio trajectories. Here we apply a deep potential molecular dynamics (DeePMD) approach to develop NNP for silica, which is the representative glassformer widely used as a model system for simulating network-forming liquids and glasses. We show that the use of a relatively small training dataset of high-temperature ab initio configurations is enough to fabricate NNP, which describes well both structural and dynamical properties of liquid silica. In particular, we calculate the pair correlation functions, angular distribution function, velocity autocorrelation functions, vibrational density of states, and mean-square displacement and reveal a close agreement with ab initio data. We show that NNP allows us to expand significantly the time-space scales achievable in simulations and thus calculating dynamical and transport properties with more accuracy than that for ab initio methods. We find that developed NNP allows us to describe the structure of the glassy silica with satisfactory accuracy even though no low-temperature configurations were included in the training procedure. The results obtained open up prospects for simulating structural and dynamical properties of liquids and glasses via NNP.}, - annotation = {WOS:000592521200001}, - author = {Balyakin, I. A. and Rempel, S. and Ryltsev, R. E. and Rempel, A. A.}, - date = {2020-11-23}, - doi = {10.1103/PhysRevE.102.052125}, - issn = {2470-0045}, - journaltitle = {Physical Review E}, - langid = {english}, - location = {{College Pk}}, - number = {5}, - pages = {052125}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. E}, - title = {Deep Machine Learning Interatomic Potential for Liquid Silica}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {102} -} - -@article{banaeiMachinelearningbasedInteratomicPotential2019, - abstract = {We report that the single interatomic potential, developed using Gaussian regression of data from density functional theory calculations, has high accuracy and flexibility to describe phonon transport with ab initio accuracy in two different atomistic configurations: perfect crystalline Si and crystalline Si with vacancies. The high accuracies of second- and third-order force constants from the Gaussian approximation potential (GAP) are demonstrated with phonon dispersion, Gruneisen parameter, three-phonon scattering rate, phonon-vacancy scattering rate, and thermal conductivity, all of which are very close to the results from density functional theory calculations. We also show that the widely used empirical potentials (Stillinger-Weber and Tersoff) produce much larger errors compared to the GAP. The computational cost of GAP is higher than the two empirical potentials, but five orders of magnitude lower than density functional theory calculations. Our work shows that GAP can provide a new opportunity for studying phonon transport in partially disordered crystalline phases with the high predictive power of ab initio calculation but at a feasible computational cost.}, - annotation = {WOS:000477921300002}, - author = {Banaei, Hasan and Guo, Ruiqiang and Hashemi, Amirreza and Lee, Sangyeop}, - date = {2019-07-26}, - doi = {10.1103/PhysRevMaterials.3.074603}, - issn = {2475-9953}, - journaltitle = {Physical Review Materials}, - langid = {english}, - location = {{College Pk}}, - number = {7}, - pages = {074603}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Mater.}, - title = {Machine-Learning-Based Interatomic Potential for Phonon Transport in Perfect Crystalline {{Si}} and Crystalline {{Si}} with Vacancies}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {3} -} - -@article{banjadeStructureMotifcentricLearning2021, - abstract = {Incorporation of physical principles in a machine learning (ML) architecture is a fundamental step toward the continued development of artificial intelligence for inorganic materials. As inspired by the Pauling's rule, we propose that structure motifs in inorganic crystals can serve as a central input to a machine learning framework. We demonstrated that the presence of structure motifs and their connections in a large set of crystalline compounds can be converted into unique vector representations using an unsupervised learning algorithm. To demonstrate the use of structure motif information, a motif-centric learning framework is created by combining motif information with the atom-based graph neural networks to form an atom-motif dual graph network (AMDNet), which is more accurate in predicting the electronic structures of metal oxides such as bandgaps. The work illustrates the route toward fundamental design of graph neural network learning architecture for complex materials by incorporating beyond-atom physical principles.}, - annotation = {WOS:000642456200011}, - author = {Banjade, Huta R. and Hauri, Sandro and Zhang, Shanshan and Ricci, Francesco and Gong, Weiyi and Hautier, Geoffroy and Vucetic, Slobodan and Yan, Qimin}, - date = {2021-04}, - doi = {10.1126/sciadv.abf1754}, - issn = {2375-2548}, - journaltitle = {Science Advances}, - langid = {english}, - location = {{Washington}}, - number = {17}, - pages = {eabf1754}, - publisher = {{Amer Assoc Advancement Science}}, - shortjournal = {Sci. Adv.}, - title = {Structure Motif-Centric Learning Framework for Inorganic Crystalline Systems}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {7} -} - -@article{barryVoxelizedAtomicStructure2020a, - abstract = {This paper introduces voxelized atomic structure (VASt) potentials as a machine learning (ML) framework for developing interatomic potentials. The VASt framework utilizes a voxelized representation of the atomic structure directly as the input to a convolutional neural network (CNN). This allows for high-fidelity representations of highly complex and diverse spatial arrangements of the atomic environments of interest. The CNN implicitly establishes the low-dimensional features needed to correlate each atomic neighborhood to its net atomic force. The selection of the salient features of the atomic structure (i.e., feature engineering) in the VASt framework is implicit, comprehensive, automated, scalable, and highly efficient. The calibrated convolutional layers learn the complex spatial relationships and multibody interactions that govern the physics of atomic systems with remarkable fidelity. We show that VASt potentials predict highly accurate forces on two phases of silicon carbide and the thermal conductivity of silicon over a range of isotropic strain.}, - annotation = {WOS:000589920000025}, - author = {Barry, Matthew C. and Wise, Kristopher E. and Kalidindi, Surya R. and Kumar, Satish}, - date = {2020-11-05}, - doi = {10.1021/acs.jpclett.0c02271}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {21}, - pages = {9093--9099}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - shorttitle = {Voxelized {{Atomic Structure Potentials}}}, - title = {Voxelized {{Atomic Structure Potentials}}: {{Predicting Atomic Forces}} with the {{Accuracy}} of {{Quantum Mechanics Using Convolutional Neural Networks}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{bartokMachineLearningGeneralpurpose2018, - abstract = {The success of first-principles electronic-structure calculation for predictive modeling in chemistry, solid-state physics, and materials science is constrained by the limitations on simulated length scales and timescales due to the computational cost and its scaling. Techniques based on machine-learning ideas for interpolating the Born-Oppenheimer potential energy surface without explicitly describing electrons have recently shown great promise, but accurately and efficiently fitting the physically relevant space of configurations remains a challenging goal. Here, we present a Gaussian approximation potential for silicon that achieves this milestone, accurately reproducing density-functional-theory reference results for a wide range of observable properties, including crystal, liquid, and amorphous bulk phases, as well as point, line, and plane defects. We demonstrate that this new potential enables calculations such as finite-temperature phase-boundary lines, self-diffusivity in the liquid, formation of the amorphous by slow quench, and dynamic brittle fracture, all of which are very expensive with a first-principles method. We show that the uncertainty quantification inherent to the Gaussian process regression framework gives a qualitative estimate of the potential’s accuracy for a given atomic configuration. The success of this model shows that it is indeed possible to create a useful machine-learning-based interatomic potential that comprehensively describes a material on the atomic scale and serves as a template for the development of such models in the future.}, - author = {Bartók, Albert P. and Kermode, James and Bernstein, Noam and Csányi, Gábor}, - date = {2018}, - doi = {10.1103/PhysRevX.8.041048}, - journaltitle = {Physical Review X}, - number = {4}, - pages = {041048}, - publisher = {{APS}}, - title = {Machine Learning a General-Purpose Interatomic Potential for Silicon}, - volume = {8} -} - -@article{Batra2020, - abstract = {Molecular dynamics (MD) is a powerful and popular tool for understanding the dynamical evolution of materials at the nano and mesoscopic scales. There are various flavors of MD ranging from the high fidelity albeit computationally expensive ab initio MD to relatively lower fidelity but much more efficient classical MD such as atomistic and coarse-grained models. Each of these different flavors of MD have been independently used by materials scientists to bring about breakthroughs in materials discovery and design. A significant gulf exists between the various MD flavors, each having varying levels of fidelity. The accuracy of DFT or ab initio MD is generally much higher than that of classical atomistic simulations which is higher than that of coarse-grained models. Multi-fidelity scale bridging to combine the accuracy and flexibility of ab initio MD with efficiency classical MD has been a longstanding goal. The advent of big-data analytics has brought to the forefront powerful machine learning methods that can be deployed to achieve this goal. Here, we provide our perspective on the challenges in multi-fidelity scale bridging and trace the developments leading up to the use of machine learning algorithms and data-science towards addressing this grand challenge.}, - author = {Batra, R and Materials, S Sankaranarayanan - Journal of Physics: and 2020, undefined}, - date = {2020}, - doi = {10.1088/2515-7639/ab8c2d}, - journaltitle = {iopscience.iop.org}, - pages = {31002}, - title = {Machine Learning for Multi-Fidelity Scale Bridging and Dynamical Simulations of Materials}, - url = {https://iopscience.iop.org/article/10.1088/2515-7639/ab8c2d/meta}, - volume = {3} -} - -@article{batznerSEEquivariantGraph2021, - abstract = {This work presents Neural Equivariant Interatomic Potentials (NequIP), a SE(3)-equivariant neural network approach for learning interatomic potentials from ab-initio calculations for molecular dynamics simulations. While most contemporary symmetry-aware models use invariant convolutions and only act on scalars, NequIP employs SE(3)-equivariant convolutions for interactions of geometric tensors, resulting in a more information-rich and faithful representation of atomic environments. The method achieves state-of-the-art accuracy on a challenging set of diverse molecules and materials while exhibiting remarkable data efficiency. NequIP outperforms existing models with up to three orders of magnitude fewer training data, challenging the widely held belief that deep neural networks require massive training sets. The high data efficiency of the method allows for the construction of accurate potentials using high-order quantum chemical level of theory as reference and enables high-fidelity molecular dynamics simulations over long time scales.}, - archiveprefix = {arXiv}, - author = {Batzner, Simon and Musaelian, Albert and Sun, Lixin and Geiger, Mario and Mailoa, Jonathan P. and Kornbluth, Mordechai and Molinari, Nicola and Smidt, Tess E. and Kozinsky, Boris}, - date = {2021-07-02}, - eprint = {2101.03164}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 17 pages, 7 figures}, - primaryclass = {cond-mat, physics:physics}, - title = {{{SE}}(3)-{{Equivariant Graph Neural Networks}} for {{Data}}-{{Efficient}} and {{Accurate Interatomic Potentials}}}, - url = {http://arxiv.org/abs/2101.03164}, - urldate = {2021-08-11} -} - -@article{bernsteinNovoExplorationSelfguided2019, - abstract = {Interatomic potential models based on machine learning (ML) are rapidly developing as tools for material simulations. However, because of their flexibility, they require large fitting databases that are normally created with substantial manual selection and tuning of reference configurations. Here, we show that ML potentials can be built in a largely automated fashion, exploring and fitting potential-energy surfaces from the beginning (de novo) within one and the same protocol. The key enabling step is the use of a configuration-averaged kernel metric that allows one to select the few most relevant and diverse structures at each step. The resulting potentials are accurate and robust for the wide range of configurations that occur during structure searching, despite only requiring a relatively small number of single-point DFT calculations on small unit cells. We apply the method to materials with diverse chemical nature and coordination environments, marking an important step toward the more routine application of ML potentials in physics, chemistry, and materials science.}, - annotation = {WOS:000489649400001}, - author = {Bernstein, Noam and Csanyi, Gabor and Deringer, Volker L.}, - date = {2019-10-11}, - doi = {10.1038/s41524-019-0236-6}, - issn = {2057-3960}, - journaltitle = {Npj Computational Materials}, - langid = {english}, - location = {{London}}, - pages = {99}, - publisher = {{Springernature}}, - shortjournal = {npj Comput. Mater.}, - title = {De Novo Exploration and Self-Guided Learning of Potential-Energy Surfaces}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {5} -} - -@article{bhowmikPerspectiveInverseDesign2019a, - abstract = {Understanding and controlling the complex and dynamic processes at battery interfaces holds the key to developing more durable and ultra high performance secondary batteries. Interfacial processes like dendrite and Solid Electrolyte Interphase (SEI) formation span numerous time-and length scales, and despite decades of research, their formation, composition, structure and function still pose a conundrum. Consequently, "inverse design" of high-performance interfaces and interphases like the SEI, remains an elusive dream. Here, we present a perspective and possible blueprint for a future battery research strategy to reach this ambitious goal. Semi-supervised generative deep learning models trained on all sources of available data, i.e., extensive multifidelity datasets from multi-scale computer simulations and databases, operando characterization from largescale research facilities, high-throughput synthesis and laboratory testing, need to work closely together to unlock this dream. We show how understanding and tracking different types of uncertainties in the experimental and simulation methods, as well as the machine learning framework for the generative model, is crucial for controlling and improving the fidelity in the predictive design of battery interfaces and interphases. We argue that simultaneous utilization of data from multiple domains, including data from failed experiments, will play a critical role in accelerating the development of reliable generative models to enable accelerated discovery and inverse design of durable ultra high performance batteries based on novel materials, structures and designs.}, - annotation = {WOS:000484341600041}, - author = {Bhowmik, Arghya and Castelli, Ivano E. and Garcia-Lastra, Juan Maria and Jorgensen, Peter Bjorn and Winther, Ole and Vegge, Tejs}, - date = {2019-09}, - doi = {10.1016/j.ensm.2019.06.011}, - issn = {2405-8297}, - journaltitle = {Energy Storage Materials}, - langid = {english}, - location = {{Amsterdam}}, - pages = {446--456}, - publisher = {{Elsevier}}, - shortjournal = {Energy Storage Mater.}, - title = {A Perspective on Inverse Design of Battery Interphases Using Multi-Scale Modelling, Experiments and Generative Deep Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {21} -} - -@article{binagiaEfficientSamplingEquilibrium, - abstract = {Generating independent configurations sampled from the physical Boltzmann distribution is an extremely difficult task, owing to the incredibly small volume in configuration space that high probability states (e.g. the folded state of a protein) occupy. Consequently, classical sampling of such Boltzmann distributions relies on the use of either Monte Carlo (MC) or Molecular Dynamics (MD) simulations to slowly propagate a valid initial configuration forward in time, which is a slow and computationally expensive process. We employ here a neural network based approach to generate statistically independent configurations of various physical models in hopes of improving sampling efficiency of such states. We first show that our trained Boltzmann generator model can quantitatively recreate an analytical solution for the free energy of a double-well potential/ We then proceed to apply our Boltzmann generator to a simple harmonic oscillator confined in a box, and show that the agreement between generated samples and the analytical result converges as training proceeds. We conclude our report with perspectives on ongoing and future work.}, - author = {Binagia, Jeremy and Friedowitz, Sean and Hou, Kevin J}, - langid = {english}, - pages = {6}, - title = {Efficient {{Sampling}} of {{Equilibrium States}} Using {{Boltzmann Generators}}} -} - -@article{bisboEfficientGlobalStructure2020, - abstract = {We propose a scheme for global optimization with first-principles energy expressions of atomistic structure. While unfolding its search, the method actively learns a surrogate model of the potential energy landscape on which it performs a number of local relaxations (exploitation) and further structural searches (exploration). Assuming Gaussian processes, deploying two separate kernel widths to better capture rough features of the energy landscape while retaining a good resolution of local minima, an acquisition function is used to decide on which of the resulting structures is the more promising and should be treated at the first-principles level. The method is demonstrated to outperform by 2 orders of magnitude a well established first-principles based evolutionary algorithm in finding surface reconstructions. Finally, global optimization with first-principles energy expressions is utilized to identify initial stages of the edge oxidation and oxygen intercalation of graphene sheets on the Ir(111) surface.}, - author = {Bisbo, Malthe K. and Hammer, Bjørk}, - date = {2020-02}, - doi = {10.1103/physrevlett.124.086102}, - journaltitle = {Physical Review Letters}, - number = {8}, - publisher = {{American Physical Society}}, - title = {Efficient Global Structure Optimization with a Machine-Learned Surrogate Model}, - volume = {124} -} - -@article{bogojeskiEfficientPrediction3D2018, - abstract = {The Kohn–Sham scheme of density functional theory is one of the most widely used methods to solve electronic structure problems for a vast variety of atomistic systems across different scientific fields. While the method is fast relative to other first principles methods and widely successful, the computational time needed is still not negligible, making it difficult to perform calculations for very large systems or over long time-scales. In this submission, we revisit a machine learning model capable of learning the electron density and the corresponding energy functional based on a set of training examples. It allows us to bypass solving the Kohn-Sham equations, providing a significant decrease in computation time. We specifically focus on the machine learning formulation of the Hohenberg-Kohn map and its decomposability. We give results and discuss challenges, limits and future directions.}, - archiveprefix = {arXiv}, - author = {Bogojeski, Mihail and Brockherde, Felix and Vogt-Maranto, Leslie and Li, Li and Tuckerman, Mark E. and Burke, Kieron and Müller, Klaus-Robert}, - date = {2018}, - eprint = {1811.06255}, - eprinttype = {arxiv}, - title = {Efficient Prediction of {{3D}} Electron Densities Using Machine Learning} -} - -@article{bogojeskiQuantumChemicalAccuracy2020a, - abstract = {Kohn-Sham density functional theory (DFT) is a standard tool in most branches of chemistry, but accuracies for many molecules are limited to 2-3 kcal . mol(-1) with presently-available functionals. Ab initio methods, such as coupled-cluster, routinely produce much higher accuracy, but computational costs limit their application to small molecules. In this paper, we leverage machine learning to calculate coupled-cluster energies from DFT densities, reaching quantum chemical accuracy (errors below 1 kcal . mol(-1)) on test data. Moreover, density-based Delta -learning (learning only the correction to a standard DFT calculation, termed Delta -DFT ) significantly reduces the amount of training data required, particularly when molecular symmetries are included. The robustness of Delta -DFT is highlighted by correcting "on the fly" DFT-based molecular dynamics (MD) simulations of resorcinol (C6H4(OH)(2)) to obtain MD trajectories with coupled-cluster accuracy. We conclude, therefore, that Delta -DFT facilitates running gas-phase MD simulations with quantum chemical accuracy, even for strained geometries and conformer changes where standard DFT fails. High-level ab initio quantum chemical methods carry a high computational burden, thus limiting their applicability. Here, the authors employ machine learning to generate coupled-cluster energies and forces at chemical accuracy for geometry optimization and molecular dynamics from DFT densities.}, - annotation = {WOS:000582054700002}, - author = {Bogojeski, Mihail and Vogt-Maranto, Leslie and Tuckerman, Mark E. and Mueller, Klaus-Robert and Burke, Kieron}, - date = {2020-10-16}, - doi = {10.1038/s41467-020-19093-1}, - issn = {2041-1723}, - journaltitle = {Nature Communications}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - pages = {5223}, - publisher = {{Nature Research}}, - shortjournal = {Nat. Commun.}, - title = {Quantum Chemical Accuracy from Density Functional Approximations via Machine Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{bonatiNeuralNetworksbasedVariationally2019a, - abstract = {Sampling complex free-energy surfaces is one of the main challenges of modern atomistic simulation methods. The presence of kinetic bottlenecks in such surfaces often renders a direct approach useless. A popular strategy is to identify a small number of key collective variables and to introduce a bias potential that is able to favor their fluctuations in order to accelerate sampling. Here, we propose to use machine-learning techniques in conjunction with the recent variationally enhanced sampling method [O. Valsson, M. Parrinello, Phys. Rev. Lett. 113, 090601 (2014)] in order to determine such potential. This is achieved by expressing the bias as a neural network. The parameters are determined in a variational learning scheme aimed at minimizing an appropriate functional. This required the development of a more efficient minimization technique. The expressivity of neural networks allows representing rapidly varying free-energy surfaces, removes boundary effects artifacts, and allows several collective variables to be handled.}, - annotation = {WOS:000485140300014}, - author = {Bonati, Luigi and Zhang, Yue-Yu and Parrinello, Michele}, - date = {2019-09-03}, - doi = {10.1073/pnas.1907975116}, - issn = {0027-8424}, - journaltitle = {Proceedings of the National Academy of Sciences of the United States of America}, - langid = {english}, - location = {{Washington}}, - number = {36}, - pages = {17641--17647}, - publisher = {{Natl Acad Sciences}}, - shortjournal = {Proc. Natl. Acad. Sci. U. S. A.}, - title = {Neural Networks-Based Variationally Enhanced Sampling}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {116} -} - -@article{bonatiSiliconLiquidStructure2018, - abstract = {Studying the crystallization process of silicon is a challenging task since empirical potentials are not able to reproduce well the properties of both semiconducting solid and metallic liquid. On the other hand, nucleation is a rare event that occurs in much longer timescales than those achievable by ab-initio molecular dynamics. To address this problem, we train a deep neural network potential based on a set of data generated by Metadynamics simulations using a classical potential. We show how this is an effective way to collect all the relevant data for the process of interest. In order to drive efficiently the crystallization process, we introduce a new collective variable based on the Debye structure factor. We are able to encode the long-range order information in a local variable which is better suited to describe the nucleation dynamics. The reference energies are then calculated using the SCAN exchange-correlation functional, which is able to get a better description of the bonding complexity of the Si phase diagram. Finally, we recover the free energy surface with a DFT accuracy, and we compute the thermodynamics properties near the melting point, obtaining a good agreement with experimental data. In addition, we study the early stages of the crystallization process, unveiling features of the nucleation mechanism.}, - author = {Bonati, Luigi and Parrinello, Michele}, - date = {2018}, - doi = {10.1103/PhysRevLett.121.265701}, - journaltitle = {Physical review letters}, - number = {26}, - pages = {265701}, - publisher = {{APS}}, - title = {Silicon Liquid Structure and Crystal Nucleation from Ab Initio Deep Metadynamics}, - volume = {121} -} - -@article{boulogeorgosMachineLearningNanoScale2020, - abstract = {Machine learning (ML) empowers biomedical systems with the capability to optimize their performance through modeling of the available data extremely well, without using strong assumptions about the modeled system. Especially in nanoscale biosystems, where the generated data sets are too vast and complex to mentally parse without computational assist, ML is instrumental in analyzing and extracting new insights, accelerating material and structure discoveries and designing experience as well as supporting nano-scale communications and networks. However, despite these efforts, the use of ML in nano-scale biomedical engineering remains still under-explored in certain areas and research challenges are still open in fields such as structure and material design and simulations, communications and signal processing, and bio-medicine applications. In this article, we review the existing research regarding the use of ML in nano-scale biomedical engineering. In more detail, we first identify and discuss the main challenges that can be formulated as ML problems. These challenges are classified in three main categories: structure and material design and simulation, communications and signal processing and biomedicine applications. Next, we discuss the state of the art ML methodologies that are used to countermeasure the aforementioned challenges. For each of the presented methodologies, special emphasis is given to its principles, applications and limitations. Finally, we conclude the article with insightful discussions, that reveal research gaps and highlight possible future research directions.}, - archiveprefix = {arXiv}, - author = {Boulogeorgos, Alexandros-Apostolos A. and Trevlakis, Stylianos E. and Tegos, Sotiris A. and Papanikolaou, Vasilis K. and Karagiannidis, George K.}, - date = {2020-10-21}, - eprint = {2008.02195}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 27 pages, 15 figures, 1 table, Journal paper}, - primaryclass = {cs, eess}, - title = {Machine {{Learning}} in {{Nano}}-{{Scale Biomedical Engineering}}}, - url = {http://arxiv.org/abs/2008.02195}, - urldate = {2021-08-11} -} - -@article{bourgeoisTransformingSolidstatePrecipitates2020, - abstract = {Many phase transformations associated with solid-state precipitation look structurally simple, yet, inexplicably, take place with great difficulty. A classic case of difficult phase transformations is the nucleation of strengthening precipitates in high-strength lightweight aluminium alloys. Here, using a combination of atomic-scale imaging, simulations and classical nucleation theory calculations, we investigate the nucleation of the strengthening phase theta' onto a template structure in the aluminium-copper alloy system. We show that this transformation can be promoted in samples exhibiting at least one nanoscale dimension, with extremely high nucleation rates for the strengthening phase as well as for an unexpected phase. This template-directed solid-state nucleation pathway is enabled by the large influx of surface vacancies that results from heating a nanoscale solid. Template-directed nucleation is replicated in a bulk alloy as well as under electron irradiation, implying that this difficult transformation can be facilitated under the general condition of sustained excess vacancy concentrations.}, - annotation = {WOS:000549162600025}, - author = {Bourgeois, Laure and Zhang, Yong and Zhang, Zezhong and Chen, Yiqiang and Medhekar, Nikhil}, - date = {2020-03-06}, - doi = {10.1038/s41467-020-15087-1}, - issn = {2041-1723}, - journaltitle = {Nature Communications}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - pages = {1248}, - publisher = {{Nature Research}}, - shortjournal = {Nat. Commun.}, - title = {Transforming Solid-State Precipitates via Excess Vacancies}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{bull-vulpeMBFitSoftwareInfrastructure2021, - abstract = {Many-body potential energy functions (MB-PEFs), which integrate data-driven representations of many-body short-range quantum mechanical interactions with physics-based representations of many-body polarization and long-range interactions, have recently been shown to provide high accuracy in the description of molecular interactions, from the gas to the condensed phase. Here, we present MB-Fit, a software infrastructure for the automated development of MB-PEFs for generic molecules within the TTM-nrg (“Thole-type model energy”) and MB-nrg (“many-body energy”) theoretical frameworks. Besides providing all the necessary computational tools for generating TTM-nrg and MB-nrg PEFs, MB-Fit provides a seamless interface with the MBX software, a many-body energy/force calculator for computer simulations. Given the demonstrated accuracy of the MB-PEFs, we believe that MB-Fit will enable routine, predictive computer simulations of generic (small) molecules in the gas, liquid, and solid phases, including, but not limited to, the modeling of isomeric equilibria in molecular clusters, solvation processes, molecular crystals, and phase diagrams.}, - author = {Bull-Vulpe, Ethan and Riera, Marc and Goetz, Andreas and Paesani, Francesco}, - date = {2021}, - shorttitle = {{{MB}}-{{Fit}}}, - title = {{{MB}}-{{Fit}}: {{Software Infrastructure}} for {{Data}}-{{Driven Many}}-{{Body Potential Energy Functions}}}, - url = {https://chemrxiv.org/engage/api-gateway/chemrxiv/assets/orp/resource/item/60f0b8a21053174f5340c8fc/original/mb-fit-software-infrastructure-for-data-driven-many-body-potential-energy-functions.pdf} -} - -@article{burkleDeepLearningApproachFirstPrinciples2021a, - abstract = {Large-scale first-principles transport calculations, while essential for device modeling, remain computationally demanding. To overcome this bottle neck, we combine first-principles transport calculations with machine learning-based nonlinear regression. We calculate the electronic conductance through first-principles based nonequilibrium Green's function techniques for small systems and map the transport properties onto local properties using local descriptors. We show that using the local descriptor as input features for deep learning-based nonlinear regression allows us to build a robust neural network that can predict the conductance of large systems beyond that of the current state-of-the-art first-principles calculation algorithms. Our protocol is applied to alkali metal nanowires, i.e., potassium, which have unique geometrical and electronic properties and hence nontrivial transport properties. We demonstrate that within our approach we can achieve qualitative agreement with experiment at a fraction of the computational effort as compared to the direct calculation of the transport properties using conventional first-principles methods.}, - annotation = {WOS:000652836400018}, - author = {Burkle, Marius and Perera, Umesha and Gimbert, Florian and Nakamura, Hisao and Kawata, Masaaki and Asai, Yoshihiro}, - date = {2021-04-27}, - doi = {10.1103/PhysRevLett.126.177701}, - issn = {0031-9007}, - journaltitle = {Physical Review Letters}, - langid = {english}, - location = {{College Pk}}, - number = {17}, - pages = {177701}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Lett.}, - title = {Deep-{{Learning Approach}} to {{First}}-{{Principles Transport Simulations}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {126} -} - -@article{byggmastarGaussianApproximationPotentials2020, - abstract = {We develop a set of machine-learning interatomic potentials for elemental V, Nb, Mo, Ta, and W using the Gaussian approximation potential framework. The potentials show good accuracy and transferability for elastic, thermal, liquid, defect, and surface properties. All potentials are augmented with accurate repulsive potentials, making them applicable to radiation damage simulations involving high-energy collisions. We study melting and liquid properties in detail and use the potentials to provide melting curves up to 400 GPa for all five elements.}, - annotation = {WOS:000576705400008}, - author = {Byggmastar, J. and Nordlund, K. and Djurabekova, F.}, - date = {2020-09-28}, - doi = {10.1103/PhysRevMaterials.4.093802}, - issn = {2475-9953}, - journaltitle = {Physical Review Materials}, - langid = {english}, - location = {{College Pk}}, - number = {9}, - pages = {093802}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Mater.}, - title = {Gaussian Approximation Potentials for Body-Centered-Cubic Transition Metals}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {4} -} - -@article{byggmastarMachinelearningInteratomicPotential2019a, - abstract = {We introduce a machine-learning interatomic potential for tungsten using the Gaussian approximation potential framework. We specifically focus on properties relevant for simulations of radiation-induced collision cascades and the damage they produce, including a realistic repulsive potential for the short-range many-body cascade dynamics and a good description of the liquid phase. Furthermore, the potential accurately reproduces surface properties and the energetics of vacancy and self-interstitial clusters, which have been longstanding deficiencies of existing potentials. The potential enables molecular dynamics simulations of radiation damage in tungsten with unprecedented accuracy.}, - annotation = {WOS:000490747900001}, - author = {Byggmastar, J. and Hamedani, A. and Nordlund, K. and Djurabekova, F.}, - date = {2019-10-17}, - doi = {10.1103/PhysRevB.100.144105}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {14}, - pages = {144105}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Machine-Learning Interatomic Potential for Radiation Damage and Defects in Tungsten}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {100} -} - -@article{calegariandradeStructureDisorderedMathrmTiO2020, - abstract = {Amorphous TiO2 (a-TiO2) is widely used in many fields, ranging from photoelectrochemistry to bioengineering, hence detailed knowledge of its atomic structure is of scientific and technological interest. Here we use an ab initio-based deep neural network potential (DP) to simulate large scale atomic models of crystalline and disordered TiO2 with molecular dynamics. Our DP reproduces the structural properties of all 11 TiO2 crystalline phases, predicts the densities and structure factors of molten and amorphous TiO2 with only a few percent deviation from experiments, and describes the pressure dependence of the amorphous structure in agreement with recent observations. It can be extended to model additional structures and compositions, and can be thus of great value in the study of TiO2-based nanomaterials.}, - author = {Calegari Andrade, Marcos F. and Selloni, Annabella}, - date = {2020-11-05}, - doi = {10/ghnhd5}, - journaltitle = {Physical Review Materials}, - number = {11}, - pages = {113803}, - publisher = {{American Physical Society}}, - shortjournal = {Phys. Rev. Materials}, - title = {Structure of Disordered \$\{\textbackslash mathrm\{\vphantom{\}\}}{{TiO}}\vphantom\{\}\vphantom\{\}\_\{2\}\$ Phases from Ab Initio Based Deep Neural Network Simulations}, - url = {https://link.aps.org/doi/10.1103/PhysRevMaterials.4.113803}, - urldate = {2021-08-11}, - volume = {4} -} - -@article{carboneMachineLearningXRayAbsorption2020, - abstract = {Simulations of excited state properties, such as spectral functions, are often computationally expensive and therefore not suitable for high-throughput modeling. As a proof of principle, we demonstrate that graph-based neural networks can be used to predict the x-ray absorption near-edge structure spectra of molecules to quantitative accuracy. Specifically, the predicted spectra reproduce nearly all prominent peaks, with 90\% of the predicted peak locations within 1 eVof the ground truth. Besides its own utility in spectral analysis and structure inference, our method can be combined with structure search algorithms to enable high-throughput spectrum sampling of the vast material configuration space, which opens up new pathways to material design and discovery.}, - annotation = {WOS:000526038600011}, - author = {Carbone, Matthew R. and Topsakal, Mehmet and Lu, Deyu and Yoo, Shinjae}, - date = {2020-04-16}, - doi = {10.1103/PhysRevLett.124.156401}, - issn = {0031-9007}, - journaltitle = {Physical Review Letters}, - langid = {english}, - location = {{College Pk}}, - number = {15}, - pages = {156401}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Lett.}, - title = {Machine-{{Learning X}}-{{Ray Absorption Spectra}} to {{Quantitative Accuracy}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {124} -} - -@article{chehaibouComputingRPAAdsorption2019a, - abstract = {Correlated quantum-chemical methods for condensed matter systems, such as the random phase approximation (RPA), hold the promise of reaching a level of accuracy much higher than that of conventional density functional theory approaches. However, the high computational cost of such methods hinders their broad applicability, in particular for finite- temperature molecular dynamics simulations. We propose a method that couples machine learning techniques with thermodynamic perturbation theory to estimate finite-temperature properties using correlated approximations. We apply this approach to compute the enthalpies of adsorption in zeolites and show that reliable estimates can be obtained by training a machine learning model with as few as 10 RPA energies. This approach paves the way to the broader use of computationally expensive quantum-chemical methods to predict the finite-temperature properties of condensed matter systems.}, - annotation = {WOS:000497260300043}, - author = {Chehaibou, Bilal and Badawi, Michael and Bucko, Tomas and Bazhirov, Timur and Rocca, Dario}, - date = {2019-11}, - doi = {10.1021/acs.jctc.9b00782}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {11}, - pages = {6333--6342}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {Computing {{RPA Adsorption Enthalpies}} by {{Machine Learning Thermodynamic Perturbation Theory}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {15} -} - -@article{Chen2021a, - abstract = {One contribution of 10 to a theme issue 'Topics in mathematical design of complex materials'. We present a perspective on several current research directions relevant to the mathematical design of new materials. We discuss: (i) design problems for phase-transforming and shape-morphing materials, (ii) epitaxy as an approach of central importance in the design of advanced semiconductor materials, (iii) selected design problems in soft matter, (iv) mathematical problems in magnetic materials, (v) some open problems in liquid crystals and soft materials and (vi) mathematical problems on liquid crystal colloids. The presentation combines topics from soft and hard condensed matter, with specific focus on those design themes where mathematical approaches could possibly lead to exciting progress. This article is part of the theme issue 'Topics in mathematical design of complex materials'.}, - author = {Chen, X and Fonseca, I and Ravnik, M and Slastikov, V and Zannoni, C}, - date = {2021-07}, - doi = {10.1098/rsta.2020.0108}, - journaltitle = {Philosophical transactions. Series A, Mathematical, physical, and engineering sciences}, - number = {2201}, - pages = {20200108}, - publisher = {{NLM (Medline)}}, - title = {Topics in the Mathematical Design of Materials}, - url = {https://royalsocietypublishing.org/doi/abs/10.1098/rsta.2020.0108}, - volume = {379} -} - -@article{Chen2021c, - abstract = {Machine learning has demonstrated great power in materials design, discovery, and property prediction. However, despite the success of machine learning in predicting discrete properties, challenges remain for continuous property prediction. The challenge is aggravated in crystalline solids due to crystallographic symmetry considerations and data scarcity. Here, the direct prediction of phonon density-of-states (DOS) is demonstrated using only atomic species and positions as input. Euclidean neural networks are applied, which by construction are equivariant to 3D rotations, translations, and inversion and thereby capture full crystal symmetry, and achieve high-quality prediction using a small training set of ≈ 10 3 examples with over 64 atom types. The predictive model reproduces key features of experimental data and even generalizes to materials with unseen elements, and is naturally suited to efficiently predict alloy systems without additional computational cost. The potential of the network is demonstrated by predicting a broad number of high phononic specific heat capacity materials. The work indicates an efficient approach to explore materials' phonon structure, and can further enable rapid screening for high-performance thermal storage materials and phonon-mediated superconductors. The ORCID identification number(s) for the author(s) of this article can be found under https://doi. One central objective of materials science is to establish structure-property relationships; that is, how specific atomic arrangements lead to certain macroscopic functionalities. This question is historically addressed through trial-and-error of a combination of structure and property characterization, theory, and calculation. However, recent advances in machine learning (ML) suggest a paradigm shift in how structure-property relationships can be directly constructed. [1,2] To date, ML has seen success in a growing spectrum of materials applications, including materials discovery and design, [3-6] process automation and optimization, [7,8] and prediction of materials' mechanical (elastic moduli), [9-12] thermodynamic and thermal transport (formation enthalpy, thermal conductivity, Debye temperature, heat capacity), [10,12-16] and electronic (bandgap, superconductivity, topology) properties, [11,17-24] and atomistic}, - author = {Chen, Z and Andrejevic, N and Smidt, T and Ding, Z and …, Q Xu - Advanced and 2021, undefined}, - date = {2021}, - doi = {10.1002/advs.202004214}, - journaltitle = {Wiley Online Library}, - publisher = {{John Wiley and Sons Inc}}, - title = {Direct Prediction of Phonon Density of States with Euclidean Neural Networks}, - url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/advs.202004214}, - volume = {8} -} - -@article{chenAtomicEnergiesConvolutional2018, - abstract = {Understanding interactions and structural properties at the atomic level is often a prerequisite to the design of novel materials. Theoretical studies based on quantum-mechanical first-principles calculations can provide this knowledge but at an immense computational cost. In recent years, machine learning has been successful in predicting structural properties at a much lower cost. Here we propose a simplified structure descriptor with no empirical parameters, "k-Bags", together with a scalable and comprehensive machine learning framework that can deepen our understanding of atomic properties of structures. This model can readily predict structure-energy relations that can provide results close to the accuracy of ab initio methods. The model provides chemically meaningful atomic energies enabling theoretical analysis of organic and inorganic molecular structures. Utilization of the local information provided by the atomic energies significantly improves upon the stochastic steps in our evolutionary global structure optimization, resulting in a much faster global minimum search of molecules, clusters, and surfaced supported species.}, - annotation = {WOS:000438654500047}, - author = {Chen, Xin and Jorgensen, Mathias S. and Li, Jun and Hammer, Bjork}, - date = {2018-07}, - doi = {10.1021/acs.jctc.8b00149}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {7}, - pages = {3933--3942}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {Atomic {{Energies}} from a {{Convolutional Neural Network}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/6}, - urldate = {2021-08-06}, - volume = {14} -} - -@article{Chenb, - abstract = {A detailed understanding of the material response to rapid compression is challenging and demanding. For instance, the element gold under dynamic compression exhibits complex phase transformations where there exist some large discrepancies between experimental and theoretical studies. Here, we combined large-scale molecular dynamics simulations with a deep potential to elucidate the dynamic compression processes of gold from an atomic level. The potential is constructed by accurately reproducing the free energy surfaces of density-functional-theory calculations for gold, from ambient conditions to 15 500 K and 500 GPa. Within this framework, we extend the simulations up to 200 000 atoms size, and found a much lower pressure threshold for phase transitioning from face-centered cubic (FCC) to body-centered (BCC), as compared to previous calculations. Furthermore, the transition pressure is strongly dependent on the shock direction, namely 159 GPa for ¡100¿ orientation and 219 GPa for ¡110¿ orientation, respectively. Most importantly, the accurate atomistic perspective presents that the shocked BCC structure contains unique features of medium-range and short-range orders, which is named "disorders" here. We propose a model and demonstrate that the existence of "disorders" significantly reduces the Gibbs free energies of shocked structures, therefore leading to the lowering of the phase transition pressure. The present study provides a new path to understand the structure dynamics under extreme conditions. The understanding of atomic structures under extreme conditions is of great importance in industrial applications and scientific studies, covering multidisciplinary fields of physics [1-3], chemistry [4, 5], materials [6], geophysics and astrophysics [7, 8]. Shock compression, served as a typical method to generate ultrahigh pressure-temperature (P-T) conditions [9, 10], leads to discoveries of unique thermodynamic states [11-13], unusual physical properties [14, 15], and unexpected meta-stable structures [7, 16-18]. With the development of in-situ time-resolved x-ray diffraction (XRD) methods, more attentions are paid to the phase transition process of shocked matters [17-21] rather than measuring the equation of state (EOS). Innumerous new phenomena have been observed and many experimental results present significantly different characteristics of structural transformation path between the dynamic and static compression, but the mechanism is far away to be known. There is a growing demand for theories from the atomic scale to understand the intrinsic mechanisms. Gold (Au) is a typical close-packed face-centered-cubic (FCC) structure at ambient conditions, and it can maintain FCC structure under static pressure up to several hundreds of GPa [13, 22]. Thus, many high-pressure experiments based on diamond-anvil-cell (DAC) prefer utilizing Au as calibration material [23]. Under isothermal compression at room temperature, FCC gold undergoes the transitions to hexagonal close-packed (HCP) structure at 151 to 410 GPa in theoretical calculations [24-27], and at ∼248 GPa in experiments [22]. Ab initio calculations predict that FCC to body-centered-cubic (BCC) transition would not occur until 230 GPa. For samples under dynamic compression, heating is inseparable to the compression, and the locus of possible shock states is a curve in the pressure-density-temperature space called the Hugoniot curve. In the phase diagram of Au calculated by Ref. [25], the Hugoniot curve directly crosses the melting curve instead of the FCC-BCC boundary. It means the gold only undergoes melting transition and no FCC-BCC transition would be observed under dynamic compression. Extraordinarily, due to the recent construction of high-quality in situ XRD tool, two phase transition points from FCC to BCC have been observed at 223 GPa [17] or 150∼176 GPa [18] under two different shock-compression experiments. The FCC-BCC phase boundary is deemed to exist under much higher P-T conditions [25] in theoretical calculations, largely deviating from these experimental results. This controversy indicates that the dynamic process is significantly different between the static and shock compression, causing huge disparity of phase transition pressures. Although ab initio method can precisely predict the thermodynamic properties of shocked-compressed Au, it fails to capture the dynamic response process of the atoms and the evolution of microstructures under shock compression due to the limit of simulation sizes. In addition, molecular dynamics (MD) simulations based on semiclassical interatomic potentials is difficult to describe the interatomic interactions with so wide temperature and density range. It is urgently needed to obtain the detailed response of matters under rapid compression and reveal the intrinsic mechanisms in an accurate and large-scale way. In this work, we performed MD simulations to}, - author = {Chen, B and Zeng, Q and Wang, H and Kang, D and Dai, J}, - date = {2021-07-19}, - doi = {arXiv:2006.13136}, - journaltitle = {arxiv.org}, - title = {Competitive Effect of Disorder and Defects on Dynamic Structural Transformation of Compressed Gold}, - url = {https://arxiv.org/abs/2006.13136} -} - -@article{chenCriticalReviewMachine2020a, - abstract = {Machine learning (ML) is rapidly revolutionizing many fields and is starting to change landscapes for physics and chemistry. With its ability to solve complex tasks autonomously, ML is being exploited as a radically new way to help find material correlations, understand materials chemistry, and accelerate the discovery of materials. Here, an in-depth review of the application of ML to energy materials, including rechargeable alkali-ion batteries, photovoltaics, catalysts, thermoelectrics, piezoelectrics, and superconductors, is presented. A conceptual framework is first provided for ML in materials science, with a broad overview of different ML techniques as well as best practices. This is followed by a critical discussion of how ML is applied in energy materials. This review is concluded with the perspectives on major challenges and opportunities in this exciting field.}, - annotation = {WOS:000509828200001}, - author = {Chen, Chi and Zuo, Yunxing and Ye, Weike and Li, Xiangguo and Deng, Zhi and Ong, Shyue Ping}, - date = {2020-02}, - doi = {10.1002/aenm.201903242}, - issn = {1614-6832}, - journaltitle = {Advanced Energy Materials}, - langid = {english}, - location = {{Weinheim}}, - number = {8}, - pages = {1903242}, - publisher = {{Wiley-V C H Verlag Gmbh}}, - shortjournal = {Adv. Energy Mater.}, - title = {A {{Critical Review}} of {{Machine Learning}} of {{Energy Materials}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {10} -} - -@article{Chend, - abstract = {Neutron and X-ray scattering represent two state-of-the-art materials characterization techniques that measure materials' structural and dynamical properties with high precision. These techniques play critical roles in understanding a wide variety of materials systems, from catalysis to polymers, nanomaterials to macromolecules, and energy materials to quantum materials. In recent years, neutron and X-ray scattering have received a significant boost due to the development and increased application of machine learning to materials problems. This article reviews the recent progress in applying machine learning techniques to augment various neutron and X-ray scattering techniques. We highlight the integration of machine learning methods into the typical workflow of scattering experiments. We focus on scattering problems that faced challenge with traditional methods but addressable using machine learning, such as leveraging the knowledge of simple materials to model more Page 2 of 56 complicated systems, learning with limited data or incomplete labels, identifying meaningful spectra and materials' representations for learning tasks, mitigating spectral noise, and many others. We present an outlook on a few emerging roles machine learning may play in broad types of scattering and spectroscopic problems in the foreseeable future.}, - author = {Chen, Z and Andrejevic, N and Drucker, N and Nguyen, T}, - journaltitle = {arxiv.org}, - title = {Machine Learning on Neutron and X-Ray Scattering}, - url = {https://arxiv.org/abs/2102.03024} -} - -@article{chenDeePKSComprehensiveDataDriven2021, - abstract = {We propose a general machine learning-based framework for building an accurate and widely applicable energy functional within the framework of generalized Kohn-Sham density functional theory. To this end, we develop a way of training self-consistent models that are capable of taking large datasets from different systems and different kinds of labels. We demonstrate that the functional that results from this training procedure gives chemically accurate predictions on energy, force, dipole, and electron density for a large class of molecules. It can be continuously improved when more and more data are available.}, - annotation = {WOS:000610984100015}, - author = {Chen, Yixiao and Zhang, Linfeng and Wang, Han and Weinan, E.}, - date = {2021-01-12}, - doi = {10.1021/acs.jctc.0c00872}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {1}, - pages = {170--181}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - shorttitle = {{{DeePKS}}}, - title = {{{DeePKS}}: {{A Comprehensive Data}}-{{Driven Approach}} toward {{Chemically Accurate Density Functional Theory}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {17} -} - -@article{chenDeePKSkitPackageDeveloping2021, - abstract = {We introduce DeePKS-kit, an open-source software package for developing machine learning based energy and density functional models. DeePKS-kit is interfaced with PyTorch, an open-source machine learning library, and PySCF, an ab initio computational chemistry program that provides simple and customized tools for developing quantum chemistry codes. It supports the DeePHF and DeePKS methods. In addition to explaining the details in the methodology and the software, we also provide an example of developing a chemically accurate model for water clusters. PROGRAM SUMMARY Program Title: DeePKS-kit Developer's repository link: https://github.com/deepmodeling/deepks-kit Licensing provisions: LGPL Programming language: Python Nature of problem: Modeling the energy and density functional in electronic structure problems with high accuracy by neural network models. Solving electronic ground state energy and charge density using the learned model. Solution method: DeePHS and DeePKS methods are implemented , interfaced with PyTorch and PySCF for neural network training and self-consistent field calculations. An iterative learning procedure is included to train the model self-consistently.}, - archiveprefix = {arXiv}, - arxivid = {2012.14615v2}, - author = {Chen, Y and Zhang, L and Wang, H}, - date = {2021}, - eprint = {2012.14615v2}, - eprinttype = {arxiv}, - journaltitle = {arxiv.org}, - title = {{{DeePKS}}-Kit: A Package for Developing Machine Learning-Based Chemically Accurate Energy and Density Functional Models}, - url = {https://arxiv.org/abs/2012.14615} -} - -@article{chenEfficientConstructionExcitedState2020, - abstract = {Recently, we have developed a multilayer energy-based fragment (MLEBF) method to describe excited states of large systems in which photochemically active and inert regions are separately treated with multiconfigurational and single-reference electronic structure method and their mutual polarization effects are naturally described within the many-body expansion framework. This MLEBF method has been demonstrated to provide highly accurate energies and gradients. In this work, we have further derived the MLEBF method with which highly accurate excited-state Hessian matrices of large systems are efficiently constructed. Moreover, in combination with recently proposed embedded atom neural network (EANN) model we have developed a machine learning (ML) accelerated MLEBF method (i.e., ML-MLEBF) in which photochemically inert region is entirely replaced with trained ML models. ML-MLEBF is found to improve computational efficiency of Hessian matrices in particular for large systems. Furthermore, both MLEBF and ML-MLEBF methods are highly parallel and exhibit low-scaling computational cost with multiple CPUs. The present developments could motivate combining various ML techniques with fragment-based electronic structure methods to explore Hessian-matrix-based excited-state properties of large systems.}, - annotation = {WOS:000550758200020}, - author = {Chen, Wen-Kai and Zhang, Yaolong and Jiang, Bin and Fang, Wei-Hai and Cui, Ganglong}, - date = {2020-07-09}, - doi = {10.1021/acs.jpca.0c04117}, - issn = {1089-5639}, - journaltitle = {Journal of Physical Chemistry A}, - langid = {english}, - location = {{Washington}}, - number = {27}, - pages = {5684--5695}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. A}, - title = {Efficient {{Construction}} of {{Excited}}-{{State Hessian Matrices}} with {{Machine Learning Accelerated Multilayer Energy}}-{{Based Fragment Method}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {124} -} - -@article{chenExploitingMachineLearning2020a, - abstract = {The excited-state dynamics of chromophores in complex environments determine a range of vital biological and energy capture processes. Time-resolved, multidimensional optical spectroscopies provide a key tool to investigate these processes. Although theory has the potential to decode these spectra in terms of the electronic and atomistic dynamics, the need for large numbers of excited-state electronic structure calculations severely limits first-principles predictions of multidimensional optical spectra for chromophores in the condensed phase. Here, we leverage the locality of chromophore excitations to develop machine learning models to predict the excited-state energy gap of chromophores in complex environments for efficiently constructing linear and multidimensional optical spectra. By analyzing the performance of these models, which span a hierarchy of physical approximations, across a range of chromophore-environment interaction strengths, we provide strategies for the construction of machine learning models that greatly accelerate the calculation of multidimensional optical spectra from first principles.}, - annotation = {WOS:000574906500014}, - author = {Chen, Michael S. and Zuehlsdorff, Tim J. and Morawietz, Tobias and Isborn, Christine M. and Markland, Thomas E.}, - date = {2020-09-17}, - doi = {10.1021/acs.jpclett.0c02168}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {18}, - pages = {7559--7568}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - title = {Exploiting {{Machine Learning}} to {{Efficiently Predict Multidimensional Optical Spectra}} in {{Complex Environments}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{chengCosegregationMgZn2020, - abstract = {Using aberration-corrected scanning transmission electron microscopy, the interface between η1-precipitates and Al matrix in the aged Al–2.1Zn–1.7Mg (at.\%) alloy has been investigated. Atomic-resolution X-ray dispersive spectroscopy mapping reveals the occurrence of co-segregation of Mg and Zn atoms in the \{002\}α habit plane of the η1(MgZn2) precipitates. Site-specific Mg atoms segregate outside the concave of an outermost zig-zag Zn-rich layer of the η1 habit plane. Zn segregation surround the segregated Mg atomic columns was also observed. Hybrid molecular dynamics (MD) and Monte Carlo (MC) atomistic simulations with the accuracy of density functional theory indicate the site-specific segregation is energetically favourable.}, - author = {Cheng, Bingqing and Zhao, Xiaojun and Zhang, Yong and Chen, Houwen and Polmear, Ian and Nie, Jian-Feng}, - date = {2020-08-01}, - doi = {10/gmgc5h}, - issn = {1359-6462}, - journaltitle = {Scripta Materialia}, - langid = {english}, - pages = {51--55}, - shortjournal = {Scripta Materialia}, - title = {Co-Segregation of {{Mg}} and {{Zn}} Atoms at the Planar Η1-Precipitate/{{Al}} Matrix Interface in an Aged {{Al}}–{{Zn}}–{{Mg}} Alloy}, - url = {https://www.sciencedirect.com/science/article/pii/S1359646220302098}, - urldate = {2021-08-11}, - volume = {185} -} - -@article{chengDeeplearningPotentialMethod2021, - abstract = {The shear viscosity of matter and efficient simulating methods in a wide range of temperatures and densities are desirable. In this study, we present the deep-learning many-body potential (the deep potential) method to reduce the computational cost of simulations for the viscosity of liquid aluminum at high temperature and high pressure with accurate results. Viscosities for densities of 2.35 g/cm(3), 2.7 g/cm(3), 3.5 g/cm(3), and 4.27 g/cm(3) and temperatures from melting points to about 50000 K are calculated. The results agree well with the experiment data at a pressure near 1 bar and are consistent with the simulation of first-principles at high pressure and high temperature. We reveal the behavior of the shear viscosity of liquid Al at a range where the current experimental results do not exist. Based on the available experimental data and newly generated simulation data, we propose a modified Enskog-Dymond theory, which can analytically calculate the viscosity of Al at this range. This research is helpful for numerous potential applications.}, - annotation = {WOS:000630443900005}, - author = {Cheng, Yuqing and Wang, Han and Wang, Shuaichuang and Gao, Xingyu and Li, Qiong and Fang, Jun and Song, Hongzhou and Chu, Weidong and Zhang, Gongmu and Song, Haifeng and Liu, Haifeng}, - date = {2021-01-01}, - doi = {10.1063/5.0036298}, - journaltitle = {Aip Advances}, - langid = {english}, - location = {{Melville}}, - number = {1}, - pages = {015043}, - publisher = {{Amer Inst Physics}}, - shortjournal = {AIP Adv.}, - title = {Deep-Learning Potential Method to Simulate Shear Viscosity of Liquid Aluminum at High Temperature and High Pressure by Molecular Dynamics}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{chenGoldSegregationImproves2021, - abstract = {Main observation and conclusion As a common electrocatalytic system, Au-Pt alloy particles are often prepared as Au-core-Pt-shell (Au@Pt) to make full use of platinum. However, Au has a strong tendency to segregate to the outer surface, leading to the redistribution of the active sites. Unfortunately, the mechanism of such reconstruction and its effect on the electrocatalytic activity have not been thoroughly discussed, largely owing to the complexity of in-situ characterization and computational modeling. Herein, by taking the 55-atom Au13Pt42 core-shell nanocluster as an example, we utilized the neural network potential at DFT level and the genetic algorithm to search the complex global configurational space. It turns out that it is thermodynamically favorable when all gold atoms are segregated to the surface and the shape of the cluster tends to change from icosahedron to a distorted amorphous structure (at a reduced core, DRC) with a unique gold distribution. Towards understanding the dynamic activity variation of oxygen reduction reaction (ORR) on this bimetallic Au@Pt system, oxygen adsorption energy calculations show that this reconstruction could not only increase the number of adsorption sites but also dramatically improve the ORR catalytic activity of each site, thus enhance the overall ORR reactivity. This article is protected by copyright. All rights reserved.}, - annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/cjoc.202100352}, - author = {Chen, Dingming and Lai, Zhuangzhuang and Zhang, Jiawei and Chen, Jianfu and Hu, Peijun and Wang, Haifeng}, - date = {2021-07-02}, - doi = {10/gmfw5g}, - issn = {1614-7065}, - journaltitle = {Chinese Journal of Chemistry}, - langid = {english}, - number = {n/a}, - shorttitle = {Gold {{Segregation Improves Electrocatalytic Activity}} of {{Icosahedron Au}}@{{Pt}} Nanocluster}, - title = {Gold {{Segregation Improves Electrocatalytic Activity}} of {{Icosahedron Au}}@{{Pt}} Nanocluster: Insights from Machine Learning}, - url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/cjoc.202100352}, - urldate = {2021-08-11}, - volume = {n/a} -} - -@article{chengRegressionClusteringImproved2019a, - abstract = {Machine learning (ML) in the representation of molecular-orbital-based (MOB) features has been shown to be an accurate and transferable approach to the prediction of post-Hartree-Fock correlation energies. Previous applications of MOB-ML employed Gaussian Process Regression (GPR), which provides good prediction accuracy with small training sets; however, the cost of GPR training scales cubically with the amount of data and becomes a computational bottleneck for large training sets. In the current work, we address this problem by introducing a clustering/regression/classification implementation of MOB-ML. In the first step, regression clustering (RC) is used to partition the training data to best fit an ensemble of linear regression (LR) models; in the second step, each cluster is regressed independently, using either LR or GPR; and in the third step, a random forest classifier (RFC) is trained for the prediction of cluster assignments based on MOB feature values. Upon inspection, RC is found to recapitulate chemically intuitive groupings of the frontier molecular orbitals, and the combined RC/LR/RFC and RC/GPR/RFC implementations of MOB-ML are found to provide good prediction accuracy with greatly reduced wall-clock training times. For a data set of thermalized (350 K) geometries of 7211 organic molecules of up to seven heavy atoms (QM7b-T), both RC/LR/RFC and RC/GPR/RFC reach chemical accuracy (1 kcal/mol prediction error) with only 300 training molecules, while providing 35000-fold and 4500-fold reductions in the wall-clock training time, respectively, compared to MOB-ML without clustering. The resulting models are also demonstrated to retain transferability for the prediction of large-molecule energies with only small-molecule training data. Finally, it is shown that capping the number of training data points per cluster leads to further improvements in prediction accuracy with negligible increases in wall-clock training time.}, - annotation = {WOS:000502688500012}, - author = {Cheng, Lixue and Kovachki, Nikola B. and Welborn, Matthew and Miller, Thomas F.}, - date = {2019-12}, - doi = {10.1021/acs.jctc.9b00884}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {12}, - pages = {6668--6677}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {Regression {{Clustering}} for {{Improved Accuracy}} and {{Training Costs}} with {{Molecular}}-{{Orbital}}-{{Based Machine Learning}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {15} -} - -@article{chenGroundStateEnergy2020, - abstract = {We introduce the deep post Hartree-Fock (DeePHF) method, a machine learning-based scheme for constructing accurate and transferable models for the ground-state energy of electronic structure problems. DeePHF predicts the energy difference between results of highly accurate models such as the coupled cluster method and low accuracy models such as the Hartree-Fock (HF) method, using the ground-state electronic orbitals as the input. It preserves all the symmetries of the original high accuracy model. The added computational cost is less than that of the reference HF or DFT and scales linearly with respect to system size. We examine the performance of DeePHF on organic molecular systems using publicly available data sets and obtain the state-of-art performance, particularly on large data sets.}, - annotation = {WOS:000569371100017}, - author = {Chen, Yixiao and Zhang, Linfeng and Wang, Han and Weinan, E.}, - date = {2020-09-03}, - doi = {10.1021/acs.jpca.0c03886}, - issn = {1089-5639}, - journaltitle = {Journal of Physical Chemistry A}, - langid = {english}, - location = {{Washington}}, - number = {35}, - pages = {7155--7165}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. A}, - title = {Ground {{State Energy Functional}} with {{Hartree}}-{{Fock Efficiency}} and {{Chemical Accuracy}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {124} -} - -@article{chengUniversalDensityMatrix2019a, - abstract = {We address the degree to which machine learning (ML) can be used to accurately and transferably predict post-Hartree-Fock correlation energies. Refined strategies for feature design and selection are presented, and the molecular-orbital-based machine learning (MOB-ML) method is applied to several test systems. Strikingly, for the second-order Moller-Plessett perturbation theory, coupled cluster with singles and doubles (CCSD), and CCSD with perturbative triples levels of theory, it is shown that the thermally accessible (350 K) potential energy surface for a single water molecule can be described to within 1 mhartree using a model that is trained from only a single reference calculation at a randomized geometry. To explore the breadth of chemical diversity that can be described, MOB-ML is also applied to a new dataset of thermalized (350 K) geometries of 7211 organic models with up to seven heavy atoms. In comparison with the previously reported Delta-ML method, MOB-ML is shown to reach chemical accuracy with threefold fewer training geometries. Finally, a transferability test in which models trained for seven-heavy-atom systems are used to predict energies for thirteen-heavy-atom systems reveals that MOB-ML reaches chemical accuracy with 36-fold fewer training calculations than Delta-ML (140 vs 5000 training calculations). Published under license by AIP Publishing.}, - annotation = {WOS:000463658900003}, - author = {Cheng, Lixue and Welborn, Matthew and Christensen, Anders S. and Miller, Thomas F.}, - date = {2019-04-07}, - doi = {10.1063/1.5088393}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {13}, - pages = {131103}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - shorttitle = {A Universal Density Matrix Functional from Molecular Orbital-Based Machine Learning}, - title = {A Universal Density Matrix Functional from Molecular Orbital-Based Machine Learning: {{Transferability}} across Organic Molecules}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {150} -} - -@article{chenIntegratingMachineLearning2019a, - abstract = {In this work we have combined machine learning techniques with our recently developed multilayer energy-based fragment method for studying excited states of large systems. The photochemically active and inert regions are separately treated with the complete active space self-consistent field method and the trained models. This method is demonstrated to provide accurate energies and gradients leading to essentially the same excited-state potential energy surfaces and nonadiabatic dynamics compared with full ab initio results. Furthermore, in conjunction with the use of machine learning models, this method is highly parallel and exhibits low-scaling computational cost. Finally, the present work could encourage the marriage of machine learning with fragment-based electronic structure methods to explore photochemistry of large systems.}, - annotation = {WOS:000503919300043}, - author = {Chen, Wen-Kai and Fang, Wei-Hai and Cui, Ganglong}, - date = {2019-12-19}, - doi = {10.1021/acs.jpclett.9b03113}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {24}, - pages = {7836--7841}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - title = {Integrating {{Machine Learning}} with the {{Multilayer Energy}}-{{Based Fragment Method}} for {{Excited States}} of {{Large Systems}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {10} -} - -@article{chenRepresentationSolutionsElliptic2021, - abstract = {Numerical solutions to high-dimensional partial differential equations (PDEs) based on neural networks have seen exciting developments. This paper derives complexity estimates of the solutions of d-dimensional second-order elliptic PDEs in the Barron space, that is a set of functions admitting the integral of certain parametric ridge function against a probability measure on the parameters. We prove under some appropriate assumptions that if the coefficients and the source term of the elliptic PDE lie in Barron spaces, then the solution of the PDE is ǫ-close with respect to the H1 norm to a Barron function. Moreover, we prove dimension-explicit bounds for the Barron norm of this approximate solution, depending at most polynomially on the dimension d of the PDE. As a direct consequence of the complexity estimates, the solution of the PDE can be approximated on any bounded domain by a two-layer neural network with respect to the H1 norm with a dimension-explicit convergence rate.}, - archiveprefix = {arXiv}, - author = {Chen, Ziang and Lu, Jianfeng and Lu, Yulong}, - date = {2021-06-14}, - eprint = {2106.07539}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {cs, math}, - title = {On the {{Representation}} of {{Solutions}} to {{Elliptic PDEs}} in {{Barron Spaces}}}, - url = {http://arxiv.org/abs/2106.07539}, - urldate = {2021-08-11} -} - -@article{chenTensorAlloyAutomaticAtomistic2020a, - abstract = {Atomistic modeling is important for studying physical and chemical properties of materials. Recently, machine learning interaction potentials have gained much more attentions as they can provide density functional theory level predictions within negligible time. The symmetry function descriptor based atomistic neural network is the most widely used model for modeling alloys. To precisely describe complex potential energy surfaces, integrating advanced metrics, such as force or virial stress, into training can be of great help. In this work, we propose a virtual-atom approach to model the total energy of symmetry function descriptors based atomistic neural network. Our approach creates the computation graph directly from atomic positions. Thus, the derivations of forces and virial can be handled by TensorFlow automatically and efficiently. The virtual atom approach with AutoGrad within TensorFlow allows for efficient training to not just energies and forces, but also virial stress. This new approach is implemented in our open-source program TensorAlloy, which supports constructing machine learning interaction potentials for both molecules and solids. The QM7 and SNAP/Ni-Mo datasets are used to demonstrate the performances of our program. (C) 2019 Elsevier B.V. All rights reserved.}, - annotation = {WOS:000528001100011}, - author = {Chen, Xin and Gao, Xing-Yu and Zhao, Ya-Fan and Lin, De-Ye and Chu, Wei-Dong and Song, Hai-Feng}, - date = {2020-05}, - doi = {10.1016/j.cpc.2019.107057}, - issn = {0010-4655}, - journaltitle = {Computer Physics Communications}, - langid = {english}, - location = {{Amsterdam}}, - pages = {107057}, - publisher = {{Elsevier}}, - shortjournal = {Comput. Phys. Commun.}, - shorttitle = {{{TensorAlloy}}}, - title = {{{TensorAlloy}}: {{An}} Automatic Atomistic Neural Network Program for Alloys}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {250} -} - -@article{chenUnsupervisedMachineLearning2020, - abstract = {Molecular dynamics (MD) simulation has been an invaluable tool for polymer nanocomposites (PNCs) research. MD simulation investigations provide massive data about the movements of PNCs beads on the micro-level. However, there are many challenges involved in correctly understanding and predicting the performance and properties of molecules from these massive MD simulation data. Traditional data-driven techniques are limited due to computational expense, required accuracy, and the availability of high-dimensional PNCs structures. Machine learning can give an effort to process massive data, but not all machine learning methods are available for all occasions; domain knowledge or manual intervention is required. Here, we aim to investigate unsupervised machine learning algorithms to discover clustering and shaping movements of PNCs beads under tension. Cluster Envelope Shaping Method (CESM) was proposed to analyze the clustering of beads and the shaping of clusters based on MD simulation data. The matrix-free PNC based elastomer simulation and Single-Chain Polymer Nanoparticles simulation, which demonstrate two types of deformations, verified the validity of this method.}, - author = {Chen, Zhudan and Li, Dazi and Wan, Haixiao and Liu, Minghui and Liu, Jun}, - date = {2020}, - doi = {10.1080/08927022.2020.1851028}, - journaltitle = {Molecular Simulation}, - publisher = {{Taylor and Francis Ltd.}}, - title = {Unsupervised Machine Learning Methods for Polymer Nanocomposites Data via Molecular Dynamics Simulation} -} - -@article{chirikiConstructingConvexEnergy2019a, - abstract = {We propose a global optimization strategy for atomistic structure determination based on two new concepts: a few-atom complementary energy landscape and atomic role models. Global optimization of costly energy expressions may be aided by performing some of the optimization on model energy landscapes. These are often based on a sum-of-atomic-contributions form that accurately reproduces every local energy minimum of the true energy expression. However, we propose that, by not including all atomic contributions, the resulting energy landscapes may become more convex, making the search for the global optimum more facile. A role model is someone we aspire to be more like; in the same vein we define the role model of an atom to be another atom whose local environment the first atom seeks to obtain itself. Basing a complementary energy landscape on the distance of some atoms from their role models in a feature space, we arrive at a useful few-atom complementary energy landscape. We show that relaxation in this landscape is an effective mutation when employed in an evolutionary algorithm used to identify the bulk cristobalite structure of SiO2 and the (1 x 4) surface reconstruction of anatase TiO2 (001).}, - annotation = {WOS:000503387500005}, - author = {Chiriki, Siva and Christiansen, Mads-Peter and Hammer, B.}, - date = {2019-12-19}, - doi = {10.1103/PhysRevB.100.235436}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {23}, - pages = {235436}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Constructing Convex Energy Landscapes for Atomistic Structure Optimization}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {100} -} - -@article{chmielaAccurateMolecularDynamics2020, - abstract = {We develop a combined machine learning (ML) and quantum mechanics approach that enables data-efficient reconstruction of flexible molecular force fields from high-level ab initio calculations, through the consideration of fundamental physical constraints. We discuss how such constraints are recovered and incorporated into ML models. Specifically, we use conservation of energy – a fundamental property of closed classical and quantum mechanical systems – to derive an efficient gradient-domain machine learning (GDML) model. The challenge of constructing conservative force fields is accomplished by learning in a Hilbert space of vectorvalued functions that obey the law of energy conservation. We proceed with the development of a multi-partite matching algorithm that enables a fully automated recovery of physically relevant point-group and fluxional symmetries from the training dataset into a symmetric variant of our model. The developed symmetric GDML (sGDML) approach is able to faithfully reproduce global force fields at the accuracy high-level ab initio methods, thus enabling sample intensive tasks like molecular dynamics simulations at that level of accuracy.}, - archiveprefix = {arXiv}, - author = {Chmiela, Stefan and Sauceda, Huziel E. and Tkatchenko, Alexandre and Müller, Klaus-Robert}, - date = {2020}, - doi = {10/gmgfsq}, - eprint = {1912.06401}, - eprinttype = {arxiv}, - langid = {english}, - pages = {129--154}, - primaryclass = {physics}, - title = {Accurate {{Molecular Dynamics Enabled}} by {{Efficient Physically}}-{{Constrained Machine Learning Approaches}}}, - url = {http://arxiv.org/abs/1912.06401}, - urldate = {2021-08-11}, - volume = {968} -} - -@article{chmielaExactMolecularDynamics2018, - abstract = {Molecular dynamics (MD) simulations employing classical force fields constitute the cornerstone of contemporary atomistic modeling in chemistry, biology, and materials science. However, the predictive power of these simulations is only as good as the underlying interatomic potential. Classical potentials often fail to faithfully capture key quantum effects in molecules and materials. Here we enable the direct construction of flexible molecular force fields from high-level ab initio calculations by incorporating spatial and temporal physical symmetries into a gradient-domain machine learning (sGDML) model in an automatic data-driven way. The developed sGDML approach faithfully reproduces global force fields at quantum-chemical CCSD(T) level of accuracy and allows converged molecular dynamics simulations with fully quantized electrons and nuclei. We present MD simulations, for flexible molecules with up to a few dozen atoms and provide insights into the dynamical behavior of these molecules. Our approach provides the key missing ingredient for achieving spectroscopic accuracy in molecular simulations.}, - annotation = {WOS:000445329000022}, - author = {Chmiela, Stefan and Sauceda, Huziel E. and Mueller, Klaus-Robert and Tkatchenko, Alexandre}, - date = {2018-09-24}, - doi = {10.1038/s41467-018-06169-2}, - issn = {2041-1723}, - journaltitle = {Nature Communications}, - langid = {english}, - location = {{London}}, - pages = {3887}, - publisher = {{Nature Publishing Group}}, - shortjournal = {Nat. Commun.}, - title = {Towards Exact Molecular Dynamics Simulations with Machine-Learned Force Fields}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/6}, - urldate = {2021-08-06}, - volume = {9} -} - -@article{chmielaSGDMLConstructingAccurate2019a, - abstract = {We present an optimized implementation of the recently proposed symmetric gradient domain machine learning (sGDML) model. The sGDML model is able to faithfully reproduce global potential energy surfaces (PES) for molecules with a few dozen atoms from a limited number of user-provided reference molecular conformations and the associated atomic forces. Here, we introduce a Python software package to reconstruct and evaluate custom sGDML force fields (FFs), without requiring in-depth knowledge about the details of the model. A user-friendly command-line interface offers assistance through the complete process of model creation, in an effort to make this novel machine learning approach accessible to broad practitioners. Our paper serves as a documentation, but also includes a practical application example of how to reconstruct and use a PBEO+MBD FF for paracetamol. Finally, we show how to interface 5GDML with the FF simulation engines ASE (Larsen et al., 2017) and i-PI (Kapil et al., 2019) to run numerical experiments, including structure optimization, classical and path integral molecular dynamics and nudged elastic band calculations. (C) 2019 The Author(s). Published by Elsevier B.V.}, - annotation = {WOS:000474312900005}, - author = {Chmiela, Stefan and Sauceda, Huziel E. and Poltavsky, Igor and Mueller, Klaus-Robert and Tkatchenko, Alexandre}, - date = {2019-07}, - doi = {10.1016/j.cpc.2019.02.007}, - issn = {0010-4655}, - journaltitle = {Computer Physics Communications}, - langid = {english}, - location = {{Amsterdam}}, - pages = {38--45}, - publisher = {{Elsevier Science Bv}}, - shortjournal = {Comput. Phys. Commun.}, - shorttitle = {{{sGDML}}}, - title = {{{sGDML}}: {{Constructing}} Accurate and Data Efficient Molecular Force Fields Using Machine Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {240} -} - -@article{choiEfficientTrainingMachine2020, - abstract = {Machine learning potentials provide an efficient and comprehensive tool to simulate large scale systems inaccessible by conventional first-principles methods still in a similar level of accuracy. One critical issue in constructing machine learning potentials is to build training data sets cost-effectively that can represent the potential energy surface in a wide range of configurations. We develop a scheme named randomized atomic-system generator to produce the training sets that widely cover the potential energy surface by combining the random sampling and structural optimization. We apply the scheme to construct the machine learning potentials for simulation of chalcogen-based phase change materials. Constructed machine-learning potentials successfully simulate the dynamics of melting and crystallization processes of binary GeTe at a level comparable to first-principles simulations. The visual analysis shows that the RAG-generated training set represents the crystallization process including the amorphous phases. From the velocity autocorrelation function obtained from the molecular-dynamics simulations, we calculate the phonon density of states to analyze the vibrational properties during crystallization.}, - author = {Choi, Young-Jae and Jhi, Seung-Hoon}, - date = {2020-10-01}, - doi = {10/gmf6kr}, - issn = {1520-6106, 1520-5207}, - journaltitle = {The Journal of Physical Chemistry B}, - langid = {english}, - number = {39}, - pages = {8704--8710}, - shortjournal = {J. Phys. Chem. B}, - title = {Efficient {{Training}} of {{Machine Learning Potentials}} by a {{Randomized Atomic}}-{{System Generator}}}, - url = {https://pubs.acs.org/doi/10.1021/acs.jpcb.0c05075}, - urldate = {2021-08-10}, - volume = {124} -} - -@article{christensenFCHLRevisitedFaster2020a, - abstract = {We introduce the FCHL19 representation for atomic environments in molecules or condensed-phase systems. Machine learning models based on FCHL19 are able to yield predictions of atomic forces and energies of query compounds with chemical accuracy on the scale of milliseconds. FCHL19 is a revision of our previous work [F. A. Faber et al., J. Chem. Phys. 148, 241717 (2018)] where the representation is discretized and the individual features are rigorously optimized using Monte Carlo optimization. Combined with a Gaussian kernel function that incorporates elemental screening, chemical accuracy is reached for energy learning on the QM7b and QM9 datasets after training for minutes and hours, respectively. The model also shows good performance for non-bonded interactions in the condensed phase for a set of water clusters with a mean absolute error (MAE) binding energy error of less than 0.1 kcal/mol/molecule after training on 3200 samples. For force learning on the MD 17 dataset, our optimized model similarly displays state-of-the-art accuracy with a regressor based on Gaussian process regression. When the revised FCHL19 representation is combined with the operator quantum machine learning regressor, forces and energies can be predicted in only a few milliseconds per atom. The model presented herein is fast and lightweight enough for use in general chemistry problems as well as molecular dynamics simulations. (C) 2020 Author(s).Y}, - annotation = {WOS:000519967800007}, - author = {Christensen, Anders S. and Bratholm, Lars A. and Faber, Felix A. and von Lilienfeld, O. Anatole}, - date = {2020-01-31}, - doi = {10.1063/1.5126701}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {4}, - options = {useprefix=true}, - pages = {044107}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - shorttitle = {{{FCHL}} Revisited}, - title = {{{FCHL}} Revisited: {{Faster}} and More Accurate Quantum Machine Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {152} -} - -@article{christiansenGaussianRepresentationImage2020, - abstract = {The success of applying machine learning to speed up structure search and improve property prediction in computational chemical physics depends critically on the representation chosen for the atomistic structure. In this work, we investigate how different image representations of two planar atomistic structures (ideal graphene and graphene with a grain boundary region) influence the ability of a reinforcement learning algorithm [the Atomistic Structure Learning Algorithm (ASLA)] to identify the structures from no prior knowledge while interacting with an electronic structure program. Compared to a one-hot encoding, we find a radial Gaussian broadening of the atomic position to be beneficial for the reinforcement learning process, which may even identify the Gaussians with the most favorable broadening hyperparameters during the structural search. Providing further image representations with angular information inspired by the smooth overlap of atomic positions method, however, is not found to cause further speedup of ASLA.}, - author = {Christiansen, Mads Peter V. and Mortensen, Henrik Lund and Meldgaard, Søren Ager and Hammer, Bjørk}, - date = {2020-07}, - doi = {10.1063/5.0015571}, - journaltitle = {Journal of Chemical Physics}, - number = {4}, - publisher = {{American Institute of Physics Inc.}}, - title = {Gaussian Representation for Image Recognition and Reinforcement Learning of Atomistic Structure}, - volume = {153} -} - -@article{coleyAutonomousDiscoveryChemical2020a, - abstract = {This two-part Review examines how automation has contributed to different aspects of discovery in the chemical sciences. In this first part, we describe a classification for discoveries of physical matter (molecules, materials, devices), processes, and models and how they are unified as search problems. We then introduce a set of questions and considerations relevant to assessing the extent of autonomy. Finally, we describe many case studies of discoveries accelerated by or resulting from computer assistance and automation from the domains of synthetic chemistry, drug discovery, inorganic chemistry, and materials science. These illustrate how rapid advancements in hardware automation and machine learning continue to transform the nature of experimentation and modeling. Part two reflects on these case studies and identifies a set of open challenges for the field.}, - annotation = {WOS:000538529000001}, - author = {Coley, Connor W. and Eyke, Natalie S. and Jensen, Klavs F.}, - date = {2020-12-14}, - doi = {10.1002/anie.201909987}, - issn = {1433-7851}, - journaltitle = {Angewandte Chemie-International Edition}, - langid = {english}, - location = {{Weinheim}}, - number = {51}, - pages = {22858--22893}, - publisher = {{Wiley-V C H Verlag Gmbh}}, - shortjournal = {Angew. Chem.-Int. Edit.}, - shorttitle = {Autonomous {{Discovery}} in the {{Chemical Sciences Part I}}}, - title = {Autonomous {{Discovery}} in the {{Chemical Sciences Part I}}: {{Progress}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {59} -} - -@article{coxDielectricResponseShortranged2020, - abstract = {The dielectric nature of polar liquids underpins much of their ability to act as useful solvents, but its description is complicated by the long-ranged nature of dipolar interactions. This is particularly pronounced under the periodic boundary conditions commonly used in molecular simulations. In this article, the dielectric properties of a water model whose intermolecular electrostatic interactions are entirely short-ranged are investigated. This is done within the framework of local molecular-field theory (LMFT), which provides a well-controlled mean-field treatment of long-ranged electrostatics. This short-ranged model gives a remarkably good performance on a number of counts, and its apparent shortcomings are readily accounted for. These results not only lend support to LMFT as an approach for understanding solvation behavior, but also are relevant to those developing interaction potentials based on local descriptions of liquid structure.}, - author = {Cox, Stephen J.}, - date = {2020-08-18}, - doi = {10/ghc8bb}, - issn = {0027-8424, 1091-6490}, - journaltitle = {Proceedings of the National Academy of Sciences}, - langid = {english}, - number = {33}, - pages = {19746--19752}, - shortjournal = {Proc Natl Acad Sci USA}, - title = {Dielectric Response with Short-Ranged Electrostatics}, - url = {http://www.pnas.org/lookup/doi/10.1073/pnas.2005847117}, - urldate = {2021-08-11}, - volume = {117} -} - -@article{cruzeiroHighlyAccurateManyBody2021, - abstract = {Dinitrogen pentoxide (N2O5) is an important intermediate in the atmospheric chemistry of nitrogen oxides. Although there has been much research, the processes that govern the physical interactions between N2O5 and water are still not fully understood at a molecular level. Gaining a quantitative insight from computer simulations requires going beyond the accuracy of classical force fields while accessing length scales and time scales that are out of reach for high-level quantum-chemical approaches. To this end, we present the development of MB-nrg many-body potential energy functions for nonreactive simulations of N2O5 in water. This MB-nrg model is based on electronic structure calculations at the coupled cluster level of theory and is compatible with the successful MB-pol model for water. It provides a physically correct description of long-range many-body interactions in combination with an explicit representation of up to threebody short-range interactions in terms of multidimensional permutationally invariant polynomials. In order to further investigate the importance of the underlying interactions in the model, a TTM-nrg model was also devised. TTM-nrg is a more simplistic representation that contains only two-body short-range interactions represented through Born-Mayer functions. In this work, an active learning approach was employed to efficiently build representative training sets of monomer, dimer, and trimer structures, and benchmarks are presented to determine the accuracy of our new models in comparison to a range of density functional theory methods. By assessing the binding curves, distortion energies of N2O5, and interaction energies in clusters of N2O5 and water, we evaluate the importance of two-body and three-body short-range potentials. The results demonstrate that our MB-nrg model has high accuracy with respect to the coupled cluster reference, outperforms current density functional theory models, and thus enables highly accurate simulations of N2O5 in aqueous environments.}, - annotation = {WOS:000674289800011}, - author = {Cruzeiro, Vinicius Wilian D. and Lambros, Eleftherios and Riera, Marc and Roy, Ronak and Paesani, Francesco and Gotz, Andreas W.}, - date = {2021-07-13}, - doi = {10.1021/acs.jctc.1c00069}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {7}, - pages = {3931--3945}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - shorttitle = {Highly {{Accurate Many}}-{{Body Potentials}} for {{Simulations}} of {{N2O5}} in {{Water}}}, - title = {Highly {{Accurate Many}}-{{Body Potentials}} for {{Simulations}} of {{N2O5}} in {{Water}}: {{Benchmarks}}, {{Development}}, and {{Validation}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {17} -} - -@article{cuevas-zuviriaAnalyticalModelElectron2020a, - abstract = {We present an analytical model representation of the electron density rho(r) in molecules in the form of expansions of a few functions (exponentials and a Gaussian) per atom. Based on a former analytical model of rho(r) in atoms, we devised its molecular implementation by introducing the anisotropy inherent in the electron distribution of atoms in molecules by means of proper anisotropic functions. The resulting model named A2MD (anisotropic analytical model of density) takes an analytical form highly suitable for obtaining the electron density in large biomolecules as its computational cost scales linearly with the number of atoms. To obtain the parameters of the model, we first devised a fitting procedure to reference electron densities obtained in ab initio correlated quantum calculations. Second, in order to skip costly ab initio calculations, we also developed a machine learning (ML)-based predictor that used neural networks trained on broad molecular datasets to determine the parameters of the model. The resulting ML methodology that we named A2MDnet (A2MD network-trained) was able to provide reliable electron densities as a basis to predict molecular features without requiring quantum calculations. The results presented together with the low computational scaling associated to the A2MD representation of rho(r) suggest potential applications to obtain reliable electron densities and rho(r)-based molecular properties in biomacromolecules.}, - annotation = {WOS:000563791600015}, - author = {Cuevas-Zuviria, Bruno and Pacios, Luis F.}, - date = {2020-08-24}, - doi = {10.1021/acs.jcim.0c00197}, - issn = {1549-9596}, - journaltitle = {Journal of Chemical Information and Modeling}, - langid = {english}, - location = {{Washington}}, - number = {8}, - pages = {3831--3842}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem Inf. Model.}, - title = {Analytical {{Model}} of {{Electron Density}} and {{Its Machine Learning Inference}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {60} -} - -@article{Cui2020, - abstract = {Active learning is a branch of machine learning that deals with problems where unlabeled data is abundant yet obtaining labels is expensive. The learning algorithm has the possibility of querying a limited number of samples to obtain the corresponding labels, subsequently used for supervised learning. In this work, we consider the task of choosing the subset of samples to be labeled from a fixed finite pool of samples. We assume the pool of samples to be a random matrix and the ground truth labels to be generated by a single-layer teacher random neural network. We employ replica methods to analyze the large deviations for the accuracy achieved after supervised learning on a subset of the original pool. These large deviations then provide optimal achievable performance boundaries for any active learning algorithm. We show that the optimal learning performance can be efficiently approached by simple message-passing active learning algorithms. We also provide a comparison with the performance of some other popular active learning strategies.}, - author = {Cui, H and Saglietti, L and and Scientific, L Zdeborová - Mathematical and 2020, undefined}, - date = {2020}, - journaltitle = {proceedings.mlr.press}, - options = {useprefix=true}, - pages = {390--430}, - title = {Large Deviations for the Perceptron Model and Consequences for Active Learning}, - url = {http://proceedings.mlr.press/v107/cui20a}, - volume = {107} -} - -@article{cuiBiomolecularQMMM2021, - abstract = {QM/MM simulations have become an indispensable tool in many chemical and biochemical investigations. Considering the tremendous degree of success, including recognition by a 2013 Nobel Prize in Chemistry, are there still "burning challenges" in QM/MM methods, especially for biomolecular systems? In this short Perspective, we discuss several issues that we believe greatly impact the robustness and quantitative applicability of QM/MM simulations to many, if not all, biomolecules. We highlight these issues with observations and relevant advances from recent studies in our group and others in the field. Despite such limited scope, we hope the discussions are of general interest and will stimulate additional developments that help push the field forward in meaningful directions.}, - annotation = {WOS:000614308000001}, - author = {Cui, Qiang and Pal, Tanmoy and Xie, Luke}, - date = {2021-01-28}, - doi = {10.1021/acs.jpcb.0c09898}, - issn = {1520-6106}, - journaltitle = {Journal of Physical Chemistry B}, - langid = {english}, - location = {{Washington}}, - number = {3}, - pages = {689--702}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. B}, - shorttitle = {Biomolecular {{QM}}/{{MM Simulations}}}, - title = {Biomolecular {{QM}}/{{MM Simulations}}: {{What Are Some}} of the "{{Burning Issues}}"?}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {125} -} - -@article{daiGrainBoundaryStrengthening2020, - abstract = {Interaction between grain boundaries and impurities usually leads to significant altering of material properties. Understanding the composition-structure-property relationship of grain boundaries is a key avenue for tailoring and designing high performance materials. In this work, we studied segregation of W into ZrB2 grain boundaries by a hybrid method combining Monte Carlo (MC) and molecular dynamics (MD), and examined the effects of segregation on grain boundary strengths by MD tensile testing with a fitted machine learning potential. It is found that W prefers grain boundary sites with local compression strains due to its smaller size compared to Zr. Rich segregation patterns (including monolayer, off-center bilayer, and other complex patterns); segregation induced grain boundary structure reconstruction; and order-disorder like segregation pattern transformation are discovered. Strong segregation tendency of W into ZrB2 grain boundaries and significant improvements on grain boundary strengths are certified, which guarantees outstanding high temperature performance of ZrB2-based UHTCs.}, - annotation = {WOS:000564257400003}, - author = {Dai, Fu-Zhi and Wen, Bo and Xiang, Huimin and Zhou, Yanchun}, - date = {2020-12}, - doi = {10.1016/j.jeurceramsoc.2020.06.007}, - issn = {0955-2219}, - journaltitle = {Journal of the European Ceramic Society}, - langid = {english}, - location = {{Oxford}}, - number = {15}, - pages = {5029--5036}, - publisher = {{Elsevier Sci Ltd}}, - shortjournal = {J. Eur. Ceram. Soc.}, - shorttitle = {Grain Boundary Strengthening in {{ZrB2}} by Segregation of {{W}}}, - title = {Grain Boundary Strengthening in {{ZrB2}} by Segregation of {{W}}: {{Atomistic}} Simulations with Deep Learning Potential}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {40} -} - -@article{daiTemperatureDependentThermal2021, - abstract = {High entropy diborides are new categories of ultra-high temperature ceramics, which are believed promising candidates for applications in hypersonic vehicles. However, knowledge on high temperature thermal and mechanical properties of high entropy diborides is still lacking unit now. In this work, variations of thermal and elastic properties of high entropy (Ti0.2Zr0.2Hf0.2Nb0.2Ta0.2)B-2 with respect to temperature were predicted by molecular dynamics simulations. Firstly, a deep learning potential for Ti-Zr-Hf-Nb-Ta-B diboride system was fitted with its prediction error in energy and force respectively being 9.2 meV/atom and 208 meV/A, in comparison with first-principles calculations. Then, temperature dependent lattice constants, anisotropic thermal expansions, anisotropic phonon thermal conductivities, and elastic properties of high entropy (Ti0.2Zr0.2Hf0.2Nb0.2Ta0.2)B-2 from 0 degrees C to 2400 degrees C were evaluated, where the predicted room temperature values agree well with experimental measurements. In addition, intrinsic lattice distortions of (Ti0.2Zr0.2Hf0.2Nb0.2Ta0.2)B-2 were analyzed by displacements of atoms from their ideal positions, which are in an order of 10(-3) A and one order of magnitude smaller than those in (Ti0.2Zr0.2Hf0.2Nb0.2Ta0.2)C. It indicates that lattice distortions in (Ti0.2Zr0.2Hf0.2Nb0.2Ta0.2)B-2 is not so severe as expected. With the new paradigm of machine learning potential, deep insight into high entropy materials can be achieved in the future, since the chemical and structural complexly in high entropy materials can be well handled by machine learning potential. (C) 2021 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science \& Technology.}, - annotation = {WOS:000636043600002}, - author = {Dai, Fu-Zhi and Sun, Yinjie and Wen, Bo and Xiang, Huimin and Zhou, Yanchun}, - date = {2021-05-10}, - doi = {10.1016/j.jmst.2020.07.014}, - issn = {1005-0302}, - journaltitle = {Journal of Materials Science \& Technology}, - langid = {english}, - location = {{Shenyang}}, - pages = {8--15}, - publisher = {{Journal Mater Sci Technol}}, - shortjournal = {J. Mater. Sci. Technol.}, - shorttitle = {Temperature {{Dependent Thermal}} and {{Elastic Properties}} of {{High Entropy}} ({{Ti0}}.{{2Zr0}}.{{2Hf0}}.{{2Nb0}}.{{2Ta0}}.2){{B}}-2}, - title = {Temperature {{Dependent Thermal}} and {{Elastic Properties}} of {{High Entropy}} ({{Ti0}}.{{2Zr0}}.{{2Hf0}}.{{2Nb0}}.{{2Ta0}}.2){{B}}-2: {{Molecular Dynamics Simulation}} by {{Deep Learning Potential}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {72} -} - -@article{daiTheoreticalPredictionThermal2020, - abstract = {High entropy materials (HEMs, e.g. high entropy alloys, high entropy ceramics) have gained increasing interests due to the possibility that they can provide challenge properties unattainable by traditional materials. Though a large number of HEMs have emerged, there is still in lack of theoretical predictions and simulations on HEMs, which is probably caused by the chemical complexity of HEMs. In this work, we demonstrate that the machine learning potentials developed in recent years can overcome the complexity of HEMs, and serve as powerful theoretical tools to simulate HEMs. A deep learning potential (DLP) for high entropy (Zr0.2Hf0.2Ti0.2Nb0.2Ta0.2)C is fitted with the prediction error in energy and force being 9.4 meV/atom and 217 meV/A, respectively. The reliability and generality of the DLP are affirmed, since it can accurately predict lattice parameters and elastic constants of mono-phase carbides TMC (TM = Ti, Zr, Hf, Nb and Ta). Lattice constants (increase from 4.5707 A to 4.6727 A), thermal expansion coefficients (increase from 7.85x10(-6) K-1 to 10.58 x 10(-6) K-1), phonon thermal conductivities (decrease from 2.02 W m(-1) K-1 to 0.95 W m(-1) K-1), and elastic properties of high entropy (Zr0.2Hf0.2Ti0.2Nb0.2Ta0.2)C in temperature ranging from 0 degrees C to 2400 degrees C are predicted by molecular dynamics simulations. The predicted room temperature properties agree well with experimental measurements, indicating the high accuracy of the DLP. With introducing of machine learning potentials, many problems that are intractable by traditional methods can be handled now. It is hopeful that deep insight into HEMs can be obtained in the future by such powerful methods. (C) 2020 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science \& Technology.}, - annotation = {WOS:000522855000019}, - author = {Dai, Fu-Zhi and Wen, Bo and Sun, Yinjie and Xiang, Huimin and Zhou, Yanchun}, - date = {2020-04-15}, - doi = {10.1016/j.jmst.2020.01.005}, - issn = {1005-0302}, - journaltitle = {Journal of Materials Science \& Technology}, - langid = {english}, - location = {{Shenyang}}, - pages = {168--174}, - publisher = {{Journal Mater Sci Technol}}, - shortjournal = {J. Mater. Sci. Technol.}, - title = {Theoretical Prediction on Thermal and Mechanical Properties of High Entropy ({{Zr0}}.{{2Hf0}}.{{2Ti0}}.{{2Nb0}}.{{2Ta0}}.2){{C}} by Deep Learning Potential}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {43} -} - -@article{dengRelationshipStructureMechanical2021, - abstract = {As one of the most abundant materials on Earth, silica has been widely studied in various crystal structures and glassy states. However, in terms of molecular structure and the corresponding mechanical properties between those of ordered and disordered states, partially ordered silica states are not well-explored in the relevant literature owing to a low probability of appearance in experiments and simulations. The lack of this knowledge significantly hinders the understanding of the inherent mechanism of mechanical properties and limits the applications in many engineering fields. In this study, we present an exploration of the complex interdependent relations of the structural properties of silica over a wide range of free energy surfaces and establish a machine learning-based prediction model via high-throughput molecular dynamics simulations coupling with an enhanced sampling method. Each scale of structure information of samples with varying crystallinity is analyzed. First, descriptors of silica structures were identified and selected as inputs to a deep neural network (DNN). The results indicate that our DNN-based approach can provide an accurate prediction of bulk modulus, shear modulus, and tensile strength of silica samples. Furthermore, the generalizability of the machine learning model is verified on the prediction tasks for much larger silica systems, as well as silica quenched at varying cooling rates. Overall, the enhanced sampling method can reliably accelerate the exploration of free energy surfaces and collection of training samples, and machine learning methods are effective in generating accurate and reliable predictions of mechanical properties of materials over the free energy surface.}, - annotation = {\_eprint: https://ceramics.onlinelibrary.wiley.com/doi/pdf/10.1111/jace.17779}, - author = {Deng, Yuanpeng and Du, Tao and Li, Hui}, - date = {2021-03-11}, - doi = {10/gmfw49}, - issn = {1551-2916}, - journaltitle = {Journal of the American Ceramic Society}, - langid = {english}, - number = {8}, - pages = {3910--3920}, - title = {Relationship of Structure and Mechanical Property of Silica with Enhanced Sampling and Machine Learning}, - url = {https://ceramics.onlinelibrary.wiley.com/doi/abs/10.1111/jace.17779}, - urldate = {2021-08-11}, - volume = {104} -} - -@article{deringerGeneralpurposeMachinelearningForce2020a, - abstract = {Elemental phosphorus is attracting growing interest across fundamental and applied fields of research. However, atomistic simulations of phosphorus have remained an outstanding challenge. Here, we show that a universally applicable force field for phosphorus can be created by machine learning (ML) from a suitably chosen ensemble of quantum-mechanical results. Our model is fitted to density-functional theory plus many-body dispersion (DFT+MBD) data; its accuracy is demonstrated for the exfoliation of black and violet phosphorus (yielding monolayers of "phosphorene" and "hittorfene"); its transferability is shown for the transition between the molecular and network liquid phases. An application to a phosphorene nanoribbon on an experimentally relevant length scale exemplifies the power of accurate and flexible ML-driven force fields for next-generation materials modelling. The methodology promises new insights into phosphorus as well as other structurally complex, e.g., layered solids that are relevant in diverse areas of chemistry, physics, and materials science. Atomistic simulations of phosphorus represent a challenge due to the element's highly diverse allotropic structures. Here the authors propose a general-purpose machine-learning force field for elemental phosphorus, which can describe a broad range of relevant bulk and nanostructured allotropes.}, - annotation = {WOS:000591379900003}, - author = {Deringer, Volker L. and Caro, Miguel A. and Csanyi, Gabor}, - date = {2020-10-29}, - doi = {10.1038/s41467-020-19168-z}, - issn = {2041-1723}, - journaltitle = {Nature Communications}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - pages = {5461}, - publisher = {{Nature Research}}, - shortjournal = {Nat. Commun.}, - title = {A General-Purpose Machine-Learning Force Field for Bulk and Nanostructured Phosphorus}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{deringerModellingUnderstandingBattery2020, - abstract = {The realistic computer modelling of battery materials is an important research goal, with open questions ranging from atomic-scale structure and dynamics to macroscopic phenomena. Quantum-mechanical methods offer high accuracy and predictive power in small-scale atomistic simulations, but they quickly reach their limits when complex electrochemical systems are to be studied-for example, when structural disorder or even fully amorphous phases are present, or when reactions take place at the interface between electrodes and electrolytes. In this Perspective, it is argued that emerging machine learning based interatomic potentials are promising tools for studying battery materials on the atomistic and nanometre length scales, affording quantum-mechanical accuracy yet being many orders of magnitude faster, and thereby extending the capabilities of current battery modelling methodology. Initial applications to solid-state electrolyte and anode materials in lithium-ion batteries are highlighted, and future directions and possible synergies with experiments are discussed.}, - annotation = {WOS:000576549400001}, - author = {Deringer, Volker L.}, - date = {2020-10}, - doi = {10.1088/2515-7655/abb011}, - issn = {2515-7655}, - journaltitle = {Journal of Physics-Energy}, - langid = {english}, - location = {{Bristol}}, - number = {4}, - pages = {041003}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {J. Phys-Energy}, - title = {Modelling and Understanding Battery Materials with Machine-Learning-Driven Atomistic Simulations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {2} -} - -@article{dickLearningDensityCorrect2019, - abstract = {We propose a new molecular simulation framework that combines the transferability, robustness and chemical flexibility of an ab initio method with the accuracy and efficiency of a machine learning model. The key to achieve this mix is to use a standard density functional theory (DFT) simulation as a preprocessor for the atomic and molecular information, obtaining a good quality electronic density. General, symmetry preserving, atom-centered electronic descriptors are then built from this density to train a neural network to correct the baseline DFT energies and forces. These electronic descriptors encode much more information than local atomic environments, allowing a simple neural network to reach the accuracy required for the problem of study at a negligible additional cost. The balance between accuracy and efficiency is determined by the baseline simulation. This is shown in results where high level quantum chemical accuracy is obtained for simulations of liquid water at standard DFT cost, or where high level DFT-accuracy is achieved in simulations with a low-level baseline DFT calculation, at a significantly reduced cost.}, - archiveprefix = {arXiv}, - author = {Dick, Sebastian and Fernandez-Serra, Marivi}, - date = {2019-10-14}, - doi = {10/gmgftv}, - eprint = {1812.06572}, - eprinttype = {arxiv}, - issn = {0021-9606, 1089-7690}, - journaltitle = {The Journal of Chemical Physics}, - langid = {english}, - number = {14}, - pages = {144102}, - shortjournal = {J. Chem. Phys.}, - title = {Learning from the {{Density}} to {{Correct Total Energy}} and {{Forces}} in {{First Principle Simulations}}}, - url = {http://arxiv.org/abs/1812.06572}, - urldate = {2021-08-11}, - volume = {151} -} - -@article{dralHierarchicalMachineLearning2020a, - abstract = {We present hierarchical machine learning (hML) of highly accurate potential energy surfaces (PESs). Our scheme is based on adding predictions of multiple Delta -machine learning models trained on energies and energy corrections calculated with a hierarchy of quantum chemical methods. Our (semi-)automatic procedure determines the optimal training set size and composition of each constituent machine learning model, simultaneously minimizing the computational effort necessary to achieve the required accuracy of the hML PES. Machine learning models are built using kernel ridge regression, and training points are selected with structure-based sampling. As an illustrative example, hML is applied to a high-level ab initio CH3Cl PES and is shown to significantly reduce the computational cost of generating the PES by a factor of 100 while retaining similar levels of accuracy (errors of similar to 1 cm(-1)).}, - annotation = {WOS:000537898900006}, - author = {Dral, Pavlo O. and Owens, Alec and Dral, Alexey and Csanyi, Gabor}, - date = {2020-05-29}, - doi = {10.1063/5.0006498}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {20}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Hierarchical Machine Learning of Potential Energy Surfaces}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {152} -} - -@article{dralMLatomIntegrativePlatform2021a, - abstract = {Atomistic machine learning (AML) simulations are used in chemistry at an ever-increasing pace. A large number of AML models has been developed, but their implementations are scattered among different packages, each with its own conventions for input and output. Thus, here we give an overview of our MLatom 2 software package, which provides an integrative platform for a wide variety of AML simulations by implementing from scratch and interfacing existing software for a range of state-of-the-art models. These include kernel method-based model types such as KREG (native implementation), sGDML, and GAP-SOAP as well as neural-network-based model types such as ANI, DeepPot-SE, and PhysNet. The theoretical foundations behind these methods are overviewed too. The modular structure of MLatom allows for easy extension to more AML model types. MLatom 2 also has many other capabilities useful for AML simulations, such as the support of custom descriptors, farthest-point and structure-based sampling, hyperparameter optimization, model evaluation, and automatic learning curve generation. It can also be used for such multi-step tasks as Delta-learning, self-correction approaches, and absorption spectrum simulation within the machine-learning nuclear-ensemble approach. Several of these MLatom 2 capabilities are showcased in application examples.}, - annotation = {WOS:000659201700002}, - author = {Dral, Pavlo O. and Ge, Fuchun and Xue, Bao-Xin and Hou, Yi-Fan and Pinheiro, Max and Huang, Jianxing and Barbatti, Mario}, - date = {2021-08}, - doi = {10.1007/s41061-021-00339-5}, - issn = {2365-0869}, - journaltitle = {Topics in Current Chemistry}, - langid = {english}, - location = {{Cham}}, - number = {4}, - pages = {27}, - publisher = {{Springer International Publishing Ag}}, - shortjournal = {Top. Curr. Chem.}, - shorttitle = {{{MLatom}} 2}, - title = {{{MLatom}} 2: {{An Integrative Platform}} for {{Atomistic Machine Learning}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {379} -} - -@article{dralQuantumChemistryAge2020a, - abstract = {As the quantum chemistry (QC) community embraces machine learning (ML), the number of new methods and applications based on the combination of QC and ML is surging. In this Perspective, a view of the current state of affairs in this new and exciting research field is offered, challenges of using machine learning in quantum chemistry applications are described, and potential future developments are outlined. Specifically, examples of how machine learning is used to improve the accuracy and accelerate quantum chemical research are shown. Generalization and classification of existing techniques are provided to ease the navigation in the sea of literature and to guide researchers entering the field. The emphasis of this Perspective is on supervised machine learning.}, - annotation = {WOS:000526339000051}, - author = {Dral, Pavlo O.}, - date = {2020-03-19}, - doi = {10.1021/acs.jpclett.9b03664}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {6}, - pages = {2336--2347}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - title = {Quantum {{Chemistry}} in the {{Age}} of {{Machine Learning}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{druchokEfficientGenerationCorrection2021a, - abstract = {Efficient design and screening of the novel molecules is a major challenge in drug and material design. This paper focuses on a multi-stage pipeline, in which several deep neural network models are combined to map discrete molecular representations into continuous vector space to later generate from it new molecular structures with desired properties. Here, the Attention-based Sequence-to-Sequence model is added to "spellcheck" and correct generated structures, while the oversampling in the continuous space allows generating candidate structures with desired distribution for properties and molecular descriptors, even for a small reference datasets. We further use computer simulation to validate the desired properties in the numerical experiment. With the focus on the drug design, such a pipeline allows generating novel structures with a control of Synthetic Accessibility Score and a series of metrics that assess the drug-likeliness. Our code is available at .}, - annotation = {WOS:000617743400001}, - author = {Druchok, Maksym and Yarish, Dzvenymyra and Gurbych, Oleksandr and Maksymenko, Mykola}, - date = {2021-04-30}, - doi = {10.1002/jcc.26494}, - issn = {0192-8651}, - journaltitle = {Journal of Computational Chemistry}, - langid = {english}, - location = {{Hoboken}}, - number = {11}, - pages = {746--760}, - publisher = {{Wiley}}, - shortjournal = {J. Comput. Chem.}, - title = {Toward Efficient Generation, Correction, and Properties Control of Unique Drug-like Structures}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {42} -} - -@article{druzbickiDynamicsSpectroscopyNeutronsRecent2021, - abstract = {This work provides an up-to-date overview of recent developments in neutron spectroscopic techniques and associated computational tools to interrogate the structural properties and dynamical behavior of complex and disordered materials, with a focus on those of a soft and polymeric nature. These have and continue to pave the way for new scientific opportunities simply thought unthinkable not so long ago, and have particularly benefited from advances in high-resolution, broadband techniques spanning energy transfers from the meV to the eV. Topical areas include the identification and robust assignment of low-energy modes underpinning functionality in soft solids and supramolecular frameworks, or the quantification in the laboratory of hitherto unexplored nuclear quantum effects dictating thermodynamic properties. In addition to novel classes of materials, we also discuss recent discoveries around water and its phase diagram, which continue to surprise us. All throughout, emphasis is placed on linking these ongoing and exciting experimental and computational developments to specific scientific questions in the context of the discovery of new materials for sustainable technologies.}, - annotation = {WOS:000650724800001}, - author = {Druzbicki, Kacper and Gaboardi, Mattia and Fernandez-Alonso, Felix}, - date = {2021-05}, - doi = {10.3390/polym13091440}, - journaltitle = {Polymers}, - langid = {english}, - location = {{Basel}}, - number = {9}, - pages = {1440}, - publisher = {{Mdpi}}, - shortjournal = {Polymers}, - title = {Dynamics \& {{Spectroscopy}} with {{Neutrons}}-{{Recent Developments}} \& {{Emerging Opportunities}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {13} -} - -@article{duanDataDrivenApproachesCan2020, - abstract = {High-throughput computational screening typically employs methods (i.e., density functional theory or DFT) that can fail to describe challenging molecules, such as those with strongly correlated electronic structure. In such cases, multireference (MR) correlated wavefunction theory (WFT) would be the appropriate choice but remains more challenging to carry out and automate than single-reference (SR) WFT or DFT. Numerous diagnostics have been proposed for identifying when MR character is likely to have an effect on the predictive power of SR calculations, but conflicting conclusions about diagnostic performance have been reached on small data sets. We compute 15 MR diagnostics, ranging from affordable DFT-based to more costly MR-WFT-based diagnostics, on a set of 3165 equilibrium and distorted small organic molecules containing up to six heavy atoms. Conflicting MR character assignments and low pairwise linear correlations among diagnostics are also observed over this set. We evaluate the ability of existing diagnostics to predict the percent recovery of the correlation energy, \%E-corr. None of the DFT-based diagnostics are nearly as predictive of \%E-corr as the best WFT-based diagnostics. To overcome the limitation of this cost-accuracy trade-off, we develop machine learning (ML, i.e., kernel ridge regression) models to predict WFT-based diagnostics from a combination of DFT-based diagnostics and a new, size-independent 3D geometric representation. The ML-predicted diagnostics correlate as well with MR effects as their computed (i.e., with WFT) values, significantly improving over the DFT-based diagnostics on which the models were trained. These ML models thus provide a promising approach to improve upon DFT-based diagnostic accuracy while remaining suitably low cost for high-throughput screening.}, - annotation = {WOS:000607532300032}, - author = {Duan, Chenru and Liu, Fang and Nandy, Aditya and Kulik, Heather J.}, - date = {2020-07-14}, - doi = {10.1021/acs.jctc.0c00358}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {7}, - pages = {4373--4387}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {Data-{{Driven Approaches Can Overcome}} the {{Cost}}-{{Accuracy Trade}}-{{Off}} in {{Multireference Diagnostics}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {16} -} - -@article{duanLearningFailurePredicting2019a, - abstract = {High-throughput computational screening for chemical discovery mandates the automated and unsupervised simulation of thousands of new molecules and materials. In challenging materials spaces, such as open shell transition metal chemistry, characterization requires time-consuming first-principles simulation that often necessitates human intervention. These calculations can frequently lead to a null result, e.g., the calculation does not converge or the molecule does not stay intact during a geometry optimization. To overcome this challenge toward realizing fully automated chemical discovery in transition metal chemistry, we have developed the first machine learning models that predict the likelihood of successful simulation outcomes. We train support vector machine and artificial neural network classifiers to predict simulation outcomes (i.e., geometry optimization result and degree of {$<$} S-2 {$>$} deviation) for a chosen electronic structure method based on chemical composition. For these static models, we achieve an area under the curve of at least 0.95, minimizing computational time spent on nonproductive simulations and therefore enabling efficient chemical space exploration. We introduce a metric of model uncertainty based on the distribution of points in the latent space to systematically improve model prediction confidence. In a complementary approach, we train a convolutional neural network classification model on simulation output electronic and geometric structure time series data. This dynamic model generalizes more readily than the static classifier by becoming more predictive as input simulation length increases. Finally, we describe approaches for using these models to enable autonomous job control in transition metal complex discovery.}, - annotation = {WOS:000464475500020}, - author = {Duan, Chenru and Janet, Jon Paul and Liu, Fang and Nandy, Aditya and Kulik, Heather J.}, - date = {2019-04}, - doi = {10.1021/acs.jctc.9b00057}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {4}, - pages = {2331--2345}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - shorttitle = {Learning from {{Failure}}}, - title = {Learning from {{Failure}}: {{Predicting Electronic Structure Calculation Outcomes}} with {{Machine Learning Models}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {15} -} - -@article{Dubbeldam2019, - abstract = {Molecular simulations are an excellent tool to study adsorption and diffusion in nanoporous materials. Examples of nanoporous materials are zeolites, carbon nanotubes, clays, metal-organic frameworks (MOFs), covalent organic frameworks (COFs) and zeolitic imidazolate frameworks (ZIFs). The molecular confinement these materials offer has been exploited in adsorption and catalysis for almost 50 years. Molecular simulations have provided understanding of the underlying shape selectivity, and adsorption and diffusion effects. Much of the reliability of the modeling predictions depends on the accuracy and transferability of the force field. However, flexibility and the chemical and structural diversity of MOFs add significant challenges for engineering force fields that are able to reproduce experimentally observed structural and dynamic properties. Recent developments in design, parameterization, and implementation of force fields for MOFs and zeolites are reviewed.}, - author = {Dubbeldam, D and Walton, KS and and …, TJH Vlugt - Advanced Theory and 2019, undefined}, - date = {2019-11}, - doi = {10.1002/adts.201900135}, - journaltitle = {Wiley Online Library}, - number = {11}, - options = {useprefix=true}, - publisher = {{Wiley-VCH Verlag}}, - title = {Design, Parameterization, and Implementation of Atomic Force Fields for Adsorption in Nanoporous Materials}, - url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/adts.201900135}, - volume = {2} -} - -@article{dussonAtomicClusterExpansion2021, - abstract = {The Atomic Cluster Expansion (Drautz, Phys. Rev. B 99, 2019) provides a framework to systematically derive polynomial basis functions for approximating isometry and permutation invariant functions, particularly with an eye to modelling properties of atomistic systems. Our presentation extends the derivation by proposing a precomputation algorithm that yields immediate guarantees that a complete basis is obtained. We provide a fast recursive algorithm for efficient evaluation and illustrate its performance in numerical tests. Finally, we discuss generalisations and open challenges, particularly from a numerical stability perspective, around basis optimisation and parameter estimation, paving the way towards a comprehensive analysis of the convergence to a high-fidelity reference model.}, - archiveprefix = {arXiv}, - author = {Dusson, Genevieve and Bachmayr, Markus and Csanyi, Gabor and Drautz, Ralf and Etter, Simon and van der Oord, Cas and Ortner, Christoph}, - date = {2021-05-12}, - eprint = {1911.03550}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: Title has changed in v3}, - options = {useprefix=true}, - primaryclass = {cs, math}, - shorttitle = {Atomic {{Cluster Expansion}}}, - title = {Atomic {{Cluster Expansion}}: {{Completeness}}, {{Efficiency}} and {{Stability}}}, - url = {http://arxiv.org/abs/1911.03550}, - urldate = {2021-08-11} -} - -@article{E2020, - abstract = {In recent years, tremendous progress has been made on numerical algorithms for solving partial differential equations (PDEs) in a very high dimension, using ideas from either nonlinear (multilevel) Monte Carlo or deep learning. They are potentially free of the curse of dimensionality for many different applications and have been proven to be so in the case of some nonlinear Monte Carlo methods for nonlinear parabolic PDEs. In this paper, we review these numerical and theoretical advances. In addition to algorithms based on stochastic reformulations of the original problem, such as the multilevel Picard iteration and the Deep BSDE method, we also discuss algorithms based on the more traditional Ritz, Galerkin, and least square formulations. We hope to demonstrate to the reader that studying PDEs as well as control and variational problems in very high dimensions might very well be among the most promising new directions in mathematics and scientific computing in the near future.}, - author = {E, Weinan and Han, Jiequn and Jentzen, Arnulf and arXiv preprint ArXiv:2008.13333, A Jentzen - and 2020, undefined}, - date = {2020-08}, - journaltitle = {arxiv.org}, - options = {useprefix=true}, - title = {Algorithms for Solving High Dimensional {{PDEs}}: {{From}} Nonlinear Monte Carlo to Machine Learning}, - url = {https://arxiv.org/abs/2008.13333 http://arxiv.org/abs/2008.13333} -} - -@article{ellisAcceleratingFinitetemperatureKohnSham2021a, - abstract = {We present a numerical modeling workflow based on machine learning which reproduces the total energies produced by Kohn-Sham density functional theory (DFT) at finite electronic temperature to within chemical accuracy at negligible computational cost. Based on deep neural networks, our workflow yields the local density of states (LDOS) for a given atomic configuration. From the LDOS, spatially resolved, energy-resolved, and integrated quantities can be calculated, including the DFT total free energy, which serves as the Born-Oppenheimer potential energy surface for the atoms. We demonstrate the efficacy of this approach for both solid and liquid metals and compare results between independent and unified machine-learning models for solid and liquid aluminum. Our machine-learning density functional theory framework opens up the path towards multiscale materials modeling for matter under ambient and extreme conditions at a computational scale and cost that is unattainable with current algorithms.}, - annotation = {WOS:000671587300005}, - author = {Ellis, J. A. and Fiedler, L. and Popoola, G. A. and Modine, N. A. and Stephens, J. A. and Thompson, A. P. and Cangi, A. and Rajamanickam, S.}, - date = {2021-07-08}, - doi = {10.1103/PhysRevB.104.035120}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {3}, - pages = {035120}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Accelerating Finite-Temperature {{Kohn}}-{{Sham}} Density Functional Theory with Deep Neural Networks}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {104} -} - -@article{fanNeuroevolutionMachineLearning2021, - abstract = {We develop a neuroevolution-potential (NEP) framework for generating neural network based machine-learning potentials. They are trained using an evolutionary strategy for performing large-scale molecular dynamics (MD) simulations. A descriptor of the atomic environment is constructed based on Chebyshev and Legendre polynomials. The method is implemented in graphic processing units within the open-source GPUMD package, which can attain a computational speed over \$10\^7\$ atom-step per second using one Nvidia Tesla V100. Furthermore, per-atom heat current is available in NEP, which paves the way for efficient and accurate MD simulations of heat transport in materials with strong phonon anharmonicity or spatial disorder, which usually cannot be accurately treated either with traditional empirical potentials or with perturbative methods.}, - archiveprefix = {arXiv}, - author = {Fan, Zheyong and Zeng, Zezhu and Zhang, Cunzhi and Wang, Yanzhou and Dong, Haikuan and Chen, Yue and Ala-Nissila, Tapio}, - date = {2021-07-16}, - eprint = {2107.08119}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 15 pages, 7 figures, 2 tables}, - primaryclass = {physics}, - shorttitle = {Neuroevolution Machine Learning Potentials}, - title = {Neuroevolution Machine Learning Potentials: {{Combining}} High Accuracy and Low Cost in Atomistic Simulations and Application to Heat Transport}, - url = {http://arxiv.org/abs/2107.08119}, - urldate = {2021-08-11} -} - -@article{gaiMathematicalPrincipleDeep2021, - archiveprefix = {arXiv}, - author = {Gai, Kuo and Zhang, Shihua}, - date = {2021}, - eprint = {2102.09235}, - eprinttype = {arxiv}, - shorttitle = {A {{Mathematical Principle}} of {{Deep Learning}}}, - title = {A {{Mathematical Principle}} of {{Deep Learning}}: {{Learn}} the {{Geodesic Curve}} in the {{Wasserstein Space}}} -} - -@article{Galib, - abstract = {Nitrogen oxides are removed from the troposphere through the reactive uptake of N 2 O 5 into aqueous aerosol. This process is thought to occur within the bulk of an aerosol, through solvation and subsequent hydrolysis. However, this perspective is difficult to reconcile with field measurements and cannot be verified directly because of the fast reaction kinetics of N 2 O 5. Here, we use molecular simulations, including reactive potentials and importance sampling, to study the uptake of N 2 O 5 into an aqueous aerosol. Rather than being mediated by the bulk, uptake is dominated by interfacial processes due to facile hydrolysis at the liquid-vapor interface and competitive reevaporation. With this molecular information, we propose an alternative interfacial reactive uptake model consistent with existing experimental observations.}, - author = {Galib, M and Limmer, DT}, - date = {2021-02-26}, - journaltitle = {science.sciencemag.org}, - title = {Reactive Uptake of {{N2O5}} by Atmospheric Aerosol Is Dominated by Interfacial Processes}, - url = {https://science.sciencemag.org/content/371/6532/921.abstract} -} - -@article{gaoDeepLearningProtein2020a, - abstract = {Deep learning is catalyzing a scientific revolution fueled by big data, accessible toolkits, and powerful computational resources, impacting many fields, including protein structural modeling. Protein structural modeling, such as predicting structure from amino acid sequence and evolutionary information, designing proteins toward desirable functionality, or predicting properties or behavior of a protein, is critical to understand and engineer biological systems at the molecular level. In this review, we summarize the recent advances in applying deep learning techniques to tackle problems in protein structural modeling and design. We dissect the emerging approaches using deep learning techniques for protein structural modeling and discuss advances and challenges that must be addressed. We argue for the central importance of structure, following the "sequence/structure/function'' paradigm. This review is directed to help both computational biologists to gain familiarity with the deep learning methods applied in protein modeling, and computer scientists to gain perspective on the biologically meaningful problems that may benefit from deep learning techniques.}, - annotation = {WOS:000653831500007}, - author = {Gao, Wenhao and Mahajan, Sai Pooja and Sulam, Jeremias and Gray, Jeffrey J.}, - date = {2020-12-11}, - doi = {10.1016/j.patter.2020.100142}, - issn = {2666-3899}, - journaltitle = {Patterns}, - langid = {english}, - location = {{Amsterdam}}, - number = {9}, - pages = {100142}, - publisher = {{Elsevier}}, - shortjournal = {Patterns}, - title = {Deep {{Learning}} in {{Protein Structural Modeling}} and {{Design}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {1} -} - -@article{gaoShortSolventModel2020a, - abstract = {Coulomb interactions play a major role in determining the thermodynamics, structure, and dynamics of condensed-phase systems, but often present significant challenges. Computer simulations usually use periodic boundary conditions to minimize corrections from finite cell boundaries but the long range of the Coulomb interactions generates significant contributions from distant periodic images of the simulation cell, usually calculated by Ewald sum techniques. This can add significant overhead to computer simulations and hampers the development of intuitive local pictures and simple analytic theory. In this paper, we present a general framework based on local molecular field theory to accurately determine the contributions from long-ranged Coulomb interactions to the potential of mean force between ionic or apolar hydrophobic solutes in dilute aqueous solutions described by standard classical point charge water models. The simplest approximation leads to a short solvent (SS) model, with truncated solvent-solvent and solute-solvent Coulomb interactions and long-ranged but screened Coulomb interactions only between charged solutes. The SS model accurately describes the interplay between strong short-ranged solute core interactions, local hydrogen-bond configurations, and long-ranged dielectric screening of distant charges, competing effects that are difficult to capture in standard implicit solvent models.}, - annotation = {WOS:000508977600017}, - author = {Gao, Ang and Remsing, Richard C. and Weeks, John D.}, - date = {2020-01-21}, - doi = {10.1073/pnas.1918981117}, - issn = {0027-8424}, - journaltitle = {Proceedings of the National Academy of Sciences of the United States of America}, - langid = {english}, - location = {{Washington}}, - number = {3}, - pages = {1293--1302}, - publisher = {{Natl Acad Sciences}}, - shortjournal = {Proc. Natl. Acad. Sci. U. S. A.}, - title = {Short Solvent Model for Ion Correlations and Hydrophobic Association}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {117} -} - -@article{gaoTorchANIFreeOpen2020a, - abstract = {This paper presents TorchANI, a PyTorch-based program for training/inference of ANI (ANAKIN-ME) deep learning models to obtain potential energy surfaces and other physical properties of molecular systems. ANI is an accurate neural network potential originally implemented using C++/CUDA in a program called NeuroChem. Compared with NeuroChem, TorchANI has a design emphasis on being lightweight, user friendly, cross platform, and easy to read and modify for fast prototyping, while allowing acceptable sacrifice on running performance. Because the computation of atomic environmental vectors and atomic neural networks are all implemented using PyTorch operators, TorchANI is able to use PyTorch's autograd engine to automatically compute analytical forces and Hessian matrices, as well as do force training without requiring any additional codes. TorchANI is open-source and freely available on GitHub: https://github.com/aiqm/torchani.}, - annotation = {WOS:000557375300011}, - author = {Gao, Xiang and Ramezanghorbani, Farhad and Isayev, Olexandr and Smith, Justin S. and Roitberg, Adrian E.}, - date = {2020-07-27}, - doi = {10.1021/acs.jcim.0c00451}, - issn = {1549-9596}, - journaltitle = {Journal of Chemical Information and Modeling}, - langid = {english}, - location = {{Washington}}, - number = {7}, - pages = {3408--3415}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem Inf. Model.}, - shorttitle = {{{TorchANI}}}, - title = {{{TorchANI}}: {{A Free}} and {{Open Source PyTorch}}-{{Based Deep Learning Implementation}} of the {{ANI Neural Network Potentials}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {60} -} - -@article{gartnerSignaturesLiquidliquidTransition2020, - abstract = {The possible existence of a metastable liquid-liquid transition (LLT) and a corresponding liquid-liquid critical point (LLCP) in supercooled liquid water remains a topic of much debate. An LLT has been rigorously proved in three empirically parametrized molecular models of water, and evidence consistent with an LLT has been reported for several other such models. In contrast, experimental proof of this phenomenon has been elusive due to rapid ice nucleation under deeply supercooled conditions. In this work, we combined density functional theory (DFT), machine learning, and molecular simulations to shed additional light on the possible existence of an LLT in water. We trained a deep neural network (DNN) model to represent the ab initio potential energy surface of water from DFT calculations using the Strongly Constrained and Appropriately Normed (SCAN) functional. We then used advanced sampling simulations in the multithermal-multibaric ensemble to efficiently explore the thermophysical properties of the DNN model. The simulation results are consistent with the existence of an LLCP, although they do not constitute a rigorous proof thereof. We fit the simulation data to a two-state equation of state to provide an estimate of the LLCP's location. These combined results-obtained from a purely first-principles approach with no empirical parameters-are strongly suggestive of the existence of an LLT, bolstering the hypothesis that water can separate into two distinct liquid forms.}, - annotation = {WOS:000580597300017}, - author = {Gartner, Thomas E. and Zhang, Linfeng and Piaggi, Pablo M. and Car, Roberto and Panagiotopoulos, Athanassios Z. and Debenedetti, Pablo G.}, - date = {2020-10-20}, - doi = {10.1073/pnas.2015440117}, - issn = {0027-8424}, - journaltitle = {Proceedings of the National Academy of Sciences of the United States of America}, - langid = {english}, - location = {{Washington}}, - number = {42}, - pages = {26040--26046}, - publisher = {{Natl Acad Sciences}}, - shortjournal = {Proc. Natl. Acad. Sci. U. S. A.}, - title = {Signatures of a Liquid-Liquid Transition in an Ab Initio Deep Neural Network Model for Water}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {117} -} - -@article{georgeCombiningPhononAccuracy2020, - abstract = {Machine learning driven interatomic potentials, including Gaussian approximation potential (GAP) models, are emerging tools for atomistic simulations. Here, we address the methodological question of how one can fit GAP models that accurately predict vibrational properties in specific regions of configuration space while retaining flexibility and transferability to others. We use an adaptive regularization of the GAP fit that scales with the absolute force magnitude on any given atom, thereby exploring the Bayesian interpretation of GAP regularization as an "expected error" and its impact on the prediction of physical properties for a material of interest. The approach enables excellent predictions of phonon modes (to within 0.1 THz-0.2 THz) for structurally diverse silicon allotropes, and it can be coupled with existing fitting databases for high transferability across different regions of configuration space, which we demonstrate for liquid and amorphous silicon. These findings and workflows are expected to be useful for GAP-driven materials modeling more generally.}, - annotation = {WOS:000556313400002}, - author = {George, Janine and Hautier, Geoffroy and Bartok, Albert P. and Csanyi, Gabor and Deringer, Volker L.}, - date = {2020-07-28}, - doi = {10.1063/5.0013826}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {4}, - pages = {044104}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Combining Phonon Accuracy with High Transferability in {{Gaussian}} Approximation Potential Models}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{goscinskiRoleFeatureSpace2021a, - abstract = {Efficient, physically-inspired descriptors of the structure and composition of molecules and materials play a key role in the application of machine-learning techniques to atomistic simulations. The proliferation of approaches, as well as the fact that each choice of features can lead to very different behavior depending on how they are used, e.g. by introducing non-linear kernels and non-Euclidean metrics to manipulate them, makes it difficult to objectively compare different methods, and to address fundamental questions on how one feature space is related to another. In this work we introduce a framework to compare different sets of descriptors, and different ways of transforming them by means of metrics and kernels, in terms of the structure of the feature space that they induce. We define diagnostic tools to determine whether alternative feature spaces contain equivalent amounts of information, and whether the common information is substantially distorted when going from one feature space to another. We compare, in particular, representations that are built in terms of n-body correlations of the atom density, quantitatively assessing the information loss associated with the use of low-order features. We also investigate the impact of different choices of basis functions and hyperparameters of the widely used SOAP and Behler-Parrinello features, and investigate how the use of non-linear kernels, and of a Wasserstein-type metric, change the structure of the feature space in comparison to a simpler linear feature space.}, - annotation = {WOS:000660866700001}, - author = {Goscinski, Alexander and Fraux, Guillaume and Imbalzano, Giulio and Ceriotti, Michele}, - date = {2021-06}, - doi = {10.1088/2632-2153/abdaf7}, - journaltitle = {Machine Learning-Science and Technology}, - langid = {english}, - location = {{Bristol}}, - number = {2}, - pages = {025028}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Mach. Learn.-Sci. Technol.}, - title = {The Role of Feature Space in Atomistic Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {2} -} - -@article{govoniCodeInteroperabilityExtends2021a, - abstract = {The functionality of many materials is critically dependent on the integration of dissimilar components and on the interfaces that arise between them. The description of such heterogeneous components requires the development and deployment of first principles methods, coupled to appropriate dynamical descriptions of matter and advanced sampling techniques, in order to capture all the relevant length and time scales of importance to the materials' performance. It is thus essential to build simple, streamlined computational schemes for the prediction and design of multiple properties of broad classes of materials, by developing interoperable codes which can be efficiently coupled to each other to perform complex tasks. We discuss the use of interoperable codes to simulate the structural and spectroscopic characterization of materials, including chemical reactions for catalysis, the description of defects for quantum information science, and heat and charge transport.}, - annotation = {WOS:000621097500001}, - author = {Govoni, Marco and Whitmer, Jonathan and de Pablo, Juan and Gygi, Francois and Galli, Giulia}, - date = {2021-02-19}, - doi = {10.1038/s41524-021-00501-z}, - journaltitle = {Npj Computational Materials}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - options = {useprefix=true}, - pages = {32}, - publisher = {{Nature Research}}, - shortjournal = {npj Comput. Mater.}, - title = {Code Interoperability Extends the Scope of Quantum Simulations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {7} -} - -@article{grisafiIncorporatingLongrangePhysics2019a, - abstract = {The most successful and popular machine learning models of atomic-scale properties derive their transferability from a locality ansatz. The properties of a large molecule or a bulk material are written as a sum over contributions that depend on the configurations within finite atom-centered environments. The obvious downside of this approach is that it cannot capture nonlocal, nonadditive effects such as those arising due to long-range electrostatics or quantum interference. We propose a solution to this problem by introducing nonlocal representations of the system, which are remapped as feature vectors that are defined locally and are equivariant in O(3). We consider, in particular, one form that has the same asymptotic behavior as the electrostatic potential. We demonstrate that this framework can capture nonlocal, long-range physics by building a model for the electrostatic energy of randomly distributed point-charges, for the unrelaxed binding curves of charged organic molecular dimers, and for the electronic dielectric response of liquid water. By combining a representation of the system that is sensitive to long-range correlations with the transferability of an atom-centered additive model, this method outperforms current state-of-the-art machine-learning schemes and provides a conceptual framework to incorporate nonlocal physics into atomistic machine learning. Published under license by AIP Publishing.}, - annotation = {WOS:000504063200009}, - author = {Grisafi, Andrea and Ceriotti, Michele}, - date = {2019-11-28}, - doi = {10.1063/1.5128375}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {20}, - pages = {204105}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Incorporating Long-Range Physics in Atomic-Scale Machine Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {151} -} - -@article{grisafiMultiscaleApproachPrediction2021, - abstract = {Electronic nearsightedness is one of the fundamental principles that governs the behavior of condensed matter and supports its description in terms of local entities such as chemical bonds. Locality also underlies the tremendous success of machine-learning schemes that predict quantum mechanical observables - such as the cohesive energy, the electron density, or a variety of response properties - as a sum of atom-centred contributions, based on a short-range representation of atomic environments. One of the main shortcomings of these approaches is their inability to capture physical effects ranging from electrostatic interactions to quantum delocalization, which have a long-range nature. Here we show how to build a multi-scale scheme that combines in the same framework local and non-local information, overcoming such limitations. We show that the simplest version of such features can be put in formal correspondence with a multipole expansion of permanent electrostatics. The data-driven nature of the model construction, however, makes this simple form suitable to tackle also different types of delocalized and collective effects. We present several examples that range from molecular physics to surface science and biophysics, demonstrating the ability of this multi-scale approach to model interactions driven by electrostatics, polarization and dispersion, as well as the cooperative behavior of dielectric response functions.}, - annotation = {WOS:000619216100010}, - author = {Grisafi, Andrea and Nigam, Jigyasa and Ceriotti, Michele}, - date = {2021-02-14}, - doi = {10.1039/d0sc04934d}, - issn = {2041-6520}, - journaltitle = {Chemical Science}, - langid = {english}, - location = {{Cambridge}}, - number = {6}, - pages = {2078--2090}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Chem. Sci.}, - title = {Multi-Scale Approach for the Prediction of Atomic Scale Properties}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {12} -} - -@article{Grubisic2021, - abstract = {We present PyXtal\textsubscript{F}F-a package based on Python programming language-for developing machine learning potentials (MLPs). The aim of PyXtal\textsubscript{F}F is to promote the application of atomistic simulations through providing several choices of atom-centered descriptors and machine learning regressions in one platform. Based on the given choice of descriptors (including the atom-centered symmetry functions, embedded atom density, SO4 bispectrum, and smooth SO3 power spectrum), PyXtal\textsubscript{F}F can train MLPs with either generalized linear regression or neural network models, by simultaneously minimizing the errors of energy/forces/stress tensors in comparison with the data from ab-initio simulations. The trained MLP model from PyXtal\textsubscript{F}F is interfaced with the Atomic Simulation Environment (ASE) package, which allows different types of lightweight simulations such as geometry optimization, molecular dynamics simulation, and physical properties prediction. Finally, we will illustrate the performance of PyXtal\textsubscript{F}F by applying it to investigate several material systems, including the bulk SiO 2 , high entropy alloy NbMoTaW, and elemental Pt for general purposes. Full documentation of PyXtal\textsubscript{F}F is available at https://pyxtal-ff.readthedocs.io.}, - author = {Grubišić, L and Hajba, M and Entropy, D Lacmanović -}, - date = {2021}, - doi = {10.1088/2632-2153/abc940}, - journaltitle = {mdpi.com}, - pages = {27001}, - title = {Deep Neural Network Model for Approximating Eigenmodes Localized by a Confining Potential}, - url = {https://www.mdpi.com/956466}, - volume = {2} -} - -@article{gubaevFinitetemperatureInterplayStructural2021, - abstract = {An active learning approach to train machine-learning interatomic potentials (moment tensor potentials) for multicomponent alloys to ab initio data is presented. Employing this approach, the disordered body-centered cubic (bcc) TiZrHfTax system with varying Ta concentration is investigated via molecular dynamics simulations. Our results show a strong interplay between elastic properties and the structural. phase stability, strongly affecting the mechanical properties. Based on these insights we systematically screen composition space for regimes where elastic constants show little or no temperature dependence (elinvar effect).}, - annotation = {WOS:000671591000001}, - author = {Gubaev, Konstantin and Ikeda, Yuji and Tasnadi, Ferenc and Neugebauer, Joerg and Shapeev, Alexander and Grabowski, Blazej and Koermann, Fritz}, - date = {2021-07-08}, - doi = {10.1103/PhysRevMaterials.5.073801}, - issn = {2475-9953}, - journaltitle = {Physical Review Materials}, - langid = {english}, - location = {{College Pk}}, - number = {7}, - pages = {073801}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Mater.}, - title = {Finite-Temperature Interplay of Structural Stability, Chemical Complexity, and Elastic Properties of Bcc Multicomponent Alloys from Ab Initio Trained Machine-Learning Potentials}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {5} -} - -@article{guglerEnumerationNovoInorganic2020a, - abstract = {Despite being attractive targets for functional materials, the discovery of transition metal complexes with high-throughput computational screening is challenged by the amount of feasible coordination numbers, spin states, or oxidation states and the potentially large sizes of ligands. To overcome these limitations, we take inspiration from organic chemistry where full enumeration of neutral, closed-shell molecules under the constraint of size has enriched discovery efforts. We design monodentate and bidentate ligands from scratch for the construction of mononuclear, octahedral transition metal complexes with up to 13 heavy atoms (i.e., metal, C, N, O, P, or S). From {$>$}11 000 theoretical ligands, we develop a heuristic score for ranking a chemically feasible 2500 ligand subset, only 71 of which were previously included in common organic molecule databases. We characterize the top 20\% of scored ligands with density functional theory (DFT) in an octahedral homoleptic ligand database (OHLDB). The OHLDB contains i) the geometry optimized structures of 1250 homoleptic octahedral complexes obtained from the enumerated pool of ligands and an open-shell transition metal (M(ii)/M(iii), M = Cr, Mn, Fe, or Co) and ii) the resulting high-spin/low-spin adiabatic electronic energy differences (Delta EH-L) obtained with hybrid DFT. Over the OHLDB, we observe structure-property (i.e., Delta EH-L) relationships different from those expected on the basis of ligand field arguments or from our prior data sets. Finally, we demonstrate how incorporating OHLDB data into artificial neural network (ANN) training improves ANN out-of-sample performance on much larger transition metal complexes.}, - annotation = {WOS:000508398900010}, - author = {Gugler, Stefan and Janet, Jon Paul and Kulik, Heather J.}, - date = {2020-01-01}, - doi = {10.1039/c9me00069k}, - issn = {2058-9689}, - journaltitle = {Molecular Systems Design \& Engineering}, - langid = {english}, - location = {{Cambridge}}, - number = {1}, - pages = {139--152}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Mol. Syst. Des. Eng.}, - title = {Enumeration of de Novo Inorganic Complexes for Chemical Discovery and Machine Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {5} -} - -@article{guHighrepetitionrateFemtosecondMidinfrared2019a, - abstract = {We demonstrate an effective method to obtain high-repetition-rate femtosecond mid-infrared (mid-IR) pulses by nonlinear optical modulation of mid-IR continuous-wave (CW) quantum and interband cascade lasers (ICLs and QCLs). In the experiment, a high-repetition-rate femtosecond ytterbium-doped fiber laser with nanojoule-level pulse energy was used as the pump source of optical parametric amplifiers to modulate and amplify the mid-IR CW laser. Near transform-limited 84 fs duration (7.3 cycles) mid-IR pulses were generated with above 200 mW average power and a repetition rate of 160 MHz. Moreover, the spectral tunability was demonstrated using CW QCL at different wavelengths. The scheme offered a simple method to produce high-repetition-rate ultrashort pulses and that can be flexibly adopted in other mid-IR regions. (C) 2019 Optical Society of America}, - annotation = {WOS:000499141000052}, - author = {Gu, Chenglin and Zuo, Zhong and Luo, Daping and Peng, Daowang and Di, Yuanfeng and Zou, Xing and Yang, Liu and Li, Wenxue}, - date = {2019-12-01}, - doi = {10.1364/OL.44.005848}, - issn = {0146-9592}, - journaltitle = {Optics Letters}, - langid = {english}, - location = {{Washington}}, - number = {23}, - pages = {5848--5851}, - publisher = {{Optical Soc Amer}}, - shortjournal = {Opt. Lett.}, - title = {High-Repetition-Rate Femtosecond Mid-Infrared Pulses Generated by Nonlinear Optical Modulation of Continuous-Wave {{QCLs}} and {{ICLs}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {44} -} - -@article{guNeuralNetworkRepresentation2021, - abstract = {Despite their rich information content, electronic structure data amassed at high volumes in ab initio molecular dynamics simulations are generally under-utilized. We introduce a transferable high-fidelity neural network representation of such data in the form of tight-binding Hamiltonians for crystalline materials. This predictive representation of ab initio electronic structure, combined with machine-learning boosted molecular dynamics, enables efficient and accurate electronic evolution and sampling. When applied to a one-dimension charge-density wave material, carbyne, we are able to compute the spectral function and optical conductivity in the canonical ensemble. The spectral functions evaluated during soliton-antisoliton pair annihilation process reveal significant renormalization of low-energy edge modes due to retarded electron-lattice coupling beyond the Born- Oppenheimer limit. The availability of an efficient and reusable surrogate model for the electronic structure dynamical system will enable calculating many interesting physical properties, paving way to previously inaccessible or challenging avenues in materials modeling.}, - archiveprefix = {arXiv}, - arxivid = {2107.02103v1}, - author = {Gu, Q and Zhang, L and Feng, J}, - date = {2021}, - eprint = {2107.02103v1}, - eprinttype = {arxiv}, - isbn = {2107.02103v1}, - journaltitle = {arxiv.org}, - title = {Neural Network Representation of Electronic Structure from Ab Initio Molecular Dynamics}, - url = {https://arxiv.org/abs/2011.13774} -} - -@article{guoBergmantypeMediumRange2019a, - abstract = {In recent years, some arguments about the existence of medium-range order (MRO) in the Zr-Rh system have been put forward. However, research on the structural features of the Zr-Rh binary alloy at the atomic level is still lacking. This study uses ab initio molecular dynamics simulations to systematically study the local structures of Zr77Rh23 from the liquid to the glassy states. Pair correlation function (PCF), coordination number (CN), Honeycutt-Anderson(HA) index, bond-angle distribution functions, and the Voronoi tessellation method are used to reveal a clear icosahedral-like configuration in the amorphous Zr77Rh23 alloy. It is noteworthy that the splitting in the second peak of the partial PDF implies the existence of a medium range order (MRO) in the Zr77Rh23 system. We obtain the local order in three-dimensional atomic density distributions by using a new atomistic cluster alignment (ACA) method. Interestingly, a Bergman-type MRO is observed in the glassy Zr77Rh23. Furthermore, the spatial distribution and connections of icosahedral-like clusters are shown to further demonstrate the MRO network. Our findings shed light on the nature of atomic local structures of amorphous Zr77Rh23 alloy and have important implications to understanding the formation of various MROs in metallic glasses. (C) 2019 Elsevier B.V. All rights reserved.}, - annotation = {WOS:000464663600078}, - author = {Guo, Y. R. and Qiao, Chong and Wang, J. J. and Shen, H. and Wang, S. Y. and Zheng, Y. X. and Zhang, R. J. and Chen, L. Y. and Su, Wan-Sheng and Wang, C. Z. and Ho, K. M.}, - date = {2019-06-25}, - doi = {10.1016/j.jallcom.2019.03.197}, - issn = {0925-8388}, - journaltitle = {Journal of Alloys and Compounds}, - langid = {english}, - location = {{Lausanne}}, - pages = {675--682}, - publisher = {{Elsevier Science Sa}}, - shortjournal = {J. Alloy. Compd.}, - title = {Bergman-Type Medium Range Order in Amorphous {{Zr77Rh23}} Alloy Studied by Ab Initio Molecular Dynamics Simulations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {790} -} - -@article{guoThermoelectricPerformanceNew2021, - abstract = {The structure predicting and structure screening of new structure SnSe are studied by quotient graph and deep learning potential. Using the quotient graph, we predict the new structure based on the feature of the experimental structure. Facing many structures generated by the quotient graph and reducing computation cost, the deep learning potential is used to optimize the structure. Through the electronic fitness function, it is found that the new structure with space group no. 64 shows a good electrical property of p-type and n-type. Based on the new structure, the n-type and p-type thermoelectric properties are studied by First-principles calculations and Boltzmann transport theory. The mechanical stability and thermodynamic stability of the new structure are checked, indicating that the new structure is stable. Due to the cyclic structure, the new structure shows large values of bulk modulus (B~\textasciitilde ~62.82~GPa), shear modulus (G~\textasciitilde ~43.78~GPa), Young’s modulus (E~\textasciitilde ~106.59~GPa), Debye temperature (θ~=~291 K), sound velocity (Va \textasciitilde 2890~m/s), and low value of Gruneisen parameter γ \textasciitilde 2.2, implying that strong Sn–Se atomic interactions and large lattice thermal conductivity. At 800 K, the figure of merit of p-type (1.83) is superior to that of n-type (1.68). The main scattering process of both charge transport and thermal transport is dominated by optical phonon~branches. This work not only demonstrates the excellent property of the new structure~but also provides insights to understand the transport performance of the cyclic structure.}, - author = {Guo, D. and Li, C. and Li, K. and Shao, B. and Chen, D. and Ma, Y. and Sun, J. and Cao, X. and Zeng, W. and Chang, X.}, - date = {2021-06-01}, - doi = {10/gmgd38}, - issn = {2468-6069}, - journaltitle = {Materials Today Energy}, - langid = {english}, - pages = {100665}, - shortjournal = {Materials Today Energy}, - title = {The Thermoelectric Performance of New Structure {{SnSe}} Studied by Quotient Graph and Deep Learning Potential}, - url = {https://www.sciencedirect.com/science/article/pii/S2468606921000307}, - urldate = {2021-08-11}, - volume = {20} -} - -@article{hajibabaeiSparseGaussianProcess2021a, - abstract = {For machine learning of interatomic potentials a scalable sparse Gaussian process regression formalism is introduced with a data-efficient on-the-fly adaptive sampling algorithm. With this approach, the computational cost is effectively reduced to those of the Bayesian linear regression methods while maintaining the appealing characteristics of the exact Gaussian process regression. As a showcase, experimental melting and glass-crystallization temperatures are reproduced for Li7P3S11, Li diffusivity is simulated, and an unchartered phase is revealed with much lower Li diffusivity which should be circumvented.}, - annotation = {WOS:000657121500002}, - author = {Hajibabaei, Amir and Myung, Chang Woo and Kim, Kwang S.}, - date = {2021-06-01}, - doi = {10.1103/PhysRevB.103.214102}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {21}, - pages = {214102}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - shorttitle = {Sparse {{Gaussian}} Process Potentials}, - title = {Sparse {{Gaussian}} Process Potentials: {{Application}} to Lithium Diffusivity in Superionic Conducting Solid Electrolytes}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {103} -} - -@article{hajinazarMAISEConstructionNeural2020, - abstract = {Module for ab initio structure evolution (MAISE) is an open-source package for materials modeling and prediction. The code's main feature is an automated generation of neural network (NN) interatomic potentials for use in global structure searches. The systematic construction of Behler-Parrinello-type NN models approximating ab initio energy and forces relies on two approaches introduced in our recent studies. An evolutionary sampling scheme for generating reference structures improves the NNs' mapping of regions visited in unconstrained searches, while a stratified training approach enables the creation of standardized NN models for multiple elements. A more flexible NN architecture proposed here expands the applicability of the stratified scheme for an arbitrary number of elements. The full workflow in the NN development is managed with a customizable 'MAISE-NET' wrapper written in Python. The global structure optimization capability in MAISE is based on an evolutionary algorithm applicable for nanopar-ticles, films, and bulk crystals. A multitribe extension of the algorithm allows for an efficient simultaneous optimization of nanoparticles in a given size range. Implemented structure analysis functions include fingerprinting with radial distribution functions and finding space groups with the SPGLIB tool. This work overviews MAISE's available features, constructed models, and confirmed predictions.}, - archiveprefix = {arXiv}, - arxivid = {2005.12131v2}, - author = {Hajinazar, S and Thorn, A and Sandoval, ED}, - date = {2020-09-15}, - eprint = {2005.12131v2}, - eprinttype = {arxiv}, - isbn = {2005.12131v2}, - journaltitle = {Elsevier}, - title = {{{MAISE}}: {{Construction}} of Neural Network Interatomic Models and Evolutionary Structure Optimization}, - url = {https://www.sciencedirect.com/science/article/pii/S0010465520303301} -} - -@article{haMachineLearningAssistedExcited2021, - abstract = {We present a machine learning-assisted excited state molecular dynamics (ML-ESMD) based on the ensemble density functional theory framework. Since we represent a diabatic Hamiltonian in terms of generalized valence bond ansatz within the state-interaction state-averaged spin-restricted ensemble-referenced Kohn-Sham (SI-SA-REKS) method, we can avoid singularities near conical intersections, which are crucial in excited state molecular dynamics simulations. We train the diabatic Hamiltonian elements and their analytical gradients with the SchNet architecture to construct machine learning models, while the phase freedom of off-diagonal elements of the Hamiltonian is cured by introducing the phase-less loss function. Our machine learning models show reasonable accuracy with mean absolute errors of similar to 0.1 kcal/mol and similar to 0.5 kcal/mol/A for the diabatic Hamiltonian elements and their gradients, respectively, for penta-2,4-dieniminium cation. Moreover, by exploiting the diabatic representation, our models can predict correct conical intersection structures and their topologies. In addition, our ML-ESMD simulations give almost identical result with a direct dynamics at the same level of theory.}, - annotation = {WOS:000634678200009}, - author = {Ha, Jong-Kwon and Kim, Kicheol and Min, Seung Kyu}, - date = {2021-02-09}, - doi = {10.1021/acs.jctc.0c01261}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {2}, - pages = {694--702}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {Machine {{Learning}}-{{Assisted Excited State Molecular Dynamics}} with the {{State}}-{{Interaction State}}-{{Averaged Spin}}-{{Restricted Ensemble}}-{{Referenced Kohn}}-{{Sham Approach}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {17} -} - -@article{hanDynamicObservationDendritic2020a, - abstract = {We report the laser-induced solid-state transformation between a periodic "approximant" and quasicrystal in the Al-Cr system during rapid quenching. Dynamic transmission electron microscopy allows us to capture in situ the dendritic growth of the metastable quasicrystals. The formation of dendrites during solid-state transformation is a rare phenomenon, which we attribute to the structural similarity between the two intermetallics. Through ab initio molecular dynamics simulations, we identify the dominant structural motif to be a 13-atom icosahedral cluster transcending the phases of matter.}, - annotation = {WOS:000587289700008}, - author = {Han, Insung and McKeown, Joseph T. and Tang, Ling and Wang, Cai-Zhuang and Parsamehr, Hadi and Xi, Zhucong and Lu, Ying-Rui and Kramer, Matthew J. and Shahani, Ashwin J.}, - date = {2020-11-06}, - doi = {10.1103/PhysRevLett.125.195503}, - issn = {0031-9007}, - journaltitle = {Physical Review Letters}, - langid = {english}, - location = {{College Pk}}, - number = {19}, - pages = {195503}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Lett.}, - title = {Dynamic {{Observation}} of {{Dendritic Quasicrystal Growth}} upon {{Laser}}-{{Induced Solid}}-{{State Transformation}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {125} -} - -@article{hanMachineLearningApproach2021a, - abstract = {A proper treatment of electron correlation effects is indispensable for accurate simulation of compounds. Various post-Hartree-Fock methods have been adopted to calculate correlation energies of chemical systems, but time complexity usually prevents their usage in a large scale. Here, we propose a density functional approximation, based on machine learning using neural networks, which can be readily employed to produce results comparable to second-order Moller-Plesset perturbation (MP2) ones for organic compounds with reduced computational cost. Various systems have been tested and the transferability across basis sets, structures, and nuclear configurations has been evaluated. Only a small number of molecules at the equilibrium structure has been needed for the training, and generally less than 5\% relative error has been achieved for structures outside the training domain and systems containing about 140 atoms. In addition, this approach has been applied to make predictions on correlation energies of nuclear configurations extracted from density functional theory-based molecular dynamics trajectories with only one or two structures as training data.}, - annotation = {WOS:000634678200016}, - author = {Han, Ruocheng and Rodriguez-Mayorga, Mauricio and Luber, Sandra}, - date = {2021-02-09}, - doi = {10.1021/acs.jctc.0c00898}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {2}, - pages = {777--790}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {A {{Machine Learning Approach}} for {{MP2 Correlation Energies}} and {{Its Application}} to {{Organic Compounds}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {17} -} - -@article{hanSolvingManyelectronSchrodinger2019, - abstract = {We introduce a new family of trial wave-functions based on deep neural networks to solve the many-electron Schrodinger equation. The Pauli exclusion principle is dealt with explicitly to ensure that the trial wave-functions are physical. The optimal trial wave-function is obtained through variational Monte Carlo and the computational cost scales quadratically with the number of electrons. The algorithm does not make use of any prior knowledge such as atomic orbitals. Yet it is able to represent accurately the ground-states of the tested systems, including He, H-2, Be, B, LiH, and a chain of 10 hydrogen atoms. This opens up new possibilities for solving large-scale many-electron Schrodinger equation. (C) 2019 Elsevier Inc. All rights reserved.}, - annotation = {WOS:000490766100035}, - author = {Han, Jiequn and Zhang, Linfeng and E, Weinan}, - date = {2019-12-15}, - doi = {10.1016/j.jcp.2019.108929}, - issn = {0021-9991}, - journaltitle = {Journal of Computational Physics}, - langid = {english}, - location = {{San Diego}}, - pages = {108929}, - publisher = {{Academic Press Inc Elsevier Science}}, - shortjournal = {J. Comput. Phys.}, - title = {Solving Many-Electron {{Schrodinger}} Equation Using Deep Neural Networks}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {399} -} - -@article{hanTrajectorybasedMachineLearning2020a, - abstract = {Ab initiomolecular dynamics (AIMD) has become a popular simulation technique but long simulation times are often hampered due to its high computational effort. Alternatively, classical molecular dynamics (MD) based on force fields may be used, which, however, has certain shortcomings compared to AIMD. In order to alleviate that situation, a trajectory-based machine learning (TrajML) approach is introduced for the construction of force fields by learning from AIMD trajectories. Only nuclear trajectories are required, which can be obtained by other methods beyond AIMD as well. We developed an easy-to-use MD machine learning package (TrajML MD) for instant modelling of the force field and system-focussed prediction of molecular configurations for MD trajectories. It consumes similar computational resources as classical MD but can simulate complex systems with a higher accuracy due to the targeted learning on the system of interest.}, - annotation = {WOS:000547403200001}, - author = {Han, R. and Luber, S.}, - date = {2020-10-17}, - doi = {10.1080/00268976.2020.1788189}, - issn = {0026-8976}, - journaltitle = {Molecular Physics}, - langid = {english}, - location = {{Abingdon}}, - number = {19-20}, - publisher = {{Taylor \& Francis Ltd}}, - shortjournal = {Mol. Phys.}, - title = {Trajectory-Based Machine Learning Method and Its Application to Molecular Dynamics}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {118} -} - -@article{hanUniformlyAccurateMachine2019, - abstract = {A framework is introduced for constructing interpretable and truly reliable reduced models for multiscale problems in situations without scale separation. Hydrodynamic approximation to the kinetic equation is used as an example to illustrate the main steps and issues involved. To this end, a set of generalized moments are constructed first to optimally represent the underlying velocity distribution. The well-known closure problem is then solved with the aim of best capturing the associated dynamics of the kinetic equation. The issue of physical constraints such as Galilean invariance is addressed and an active-learning procedure is introduced to help ensure that the dataset used is representative enough. The reduced system takes the form of a conventional moment system and works regardless of the numerical discretization used. Numerical results are presented for the BGK (Bhatnagar-GrossKrook) model and binary collision of Maxwell molecules. We demonstrate that the reduced model achieves a uniform accuracy in a wide range of Knudsen numbers spanning from the hydrodynamic limit to free molecular flow.}, - annotation = {WOS:000493720200012}, - author = {Han, Jiequn and Ma, Chao and Ma, Zheng and E, Weinan}, - date = {2019-10-29}, - doi = {10.1073/pnas.1909854116}, - issn = {0027-8424}, - journaltitle = {Proceedings of the National Academy of Sciences of the United States of America}, - langid = {english}, - location = {{Washington}}, - number = {44}, - pages = {21983--21991}, - publisher = {{Natl Acad Sciences}}, - shortjournal = {Proc. Natl. Acad. Sci. U. S. A.}, - title = {Uniformly Accurate Machine Learning-Based Hydrodynamic Models for Kinetic Equations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {116} -} - -@article{hanUniformlyAccurateMachine2019, - abstract = {A framework is introduced for constructing interpretable and truly reliable reduced models for multiscale problems in situations without scale separation. Hydrodynamic approximation to the kinetic equation is used as an example to illustrate the main steps and issues involved. To this end, a set of generalized moments are constructed first to optimally represent the underlying velocity distribution. The well-known closure problem is then solved with the aim of best capturing the associated dynamics of the kinetic equation. The issue of physical constraints such as Galilean invariance is addressed and an active-learning procedure is introduced to help ensure that the dataset used is representative enough. The reduced system takes the form of a conventional moment system and works regardless of the numerical discretization used. Numerical results are presented for the BGK (Bhatnagar-GrossKrook) model and binary collision of Maxwell molecules. We demonstrate that the reduced model achieves a uniform accuracy in a wide range of Knudsen numbers spanning from the hydrodynamic limit to free molecular flow.}, - annotation = {WOS:000493720200012}, - author = {Han, Jiequn and Ma, Chao and Ma, Zheng and E, Weinan}, - date = {2019-10-29}, - doi = {10.1073/pnas.1909854116}, - issn = {0027-8424}, - journaltitle = {Proceedings of the National Academy of Sciences of the United States of America}, - langid = {english}, - location = {{Washington}}, - number = {44}, - pages = {21983--21991}, - publisher = {{Natl Acad Sciences}}, - shortjournal = {Proc. Natl. Acad. Sci. U. S. A.}, - title = {Uniformly Accurate Machine Learning-Based Hydrodynamic Models for Kinetic Equations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {116} -} - -@article{hanUniversalApproximationSymmetric2019, - abstract = {We consider universal approximations of symmetric and anti-symmetric functions, which are important for applications in quantum physics, as well as other scientific and engineering computations. We give constructive approximations with explicit bounds on the number of parameters with respect to the dimension and the target accuracy ǫ. While the approximation still suffers from curse of dimensionality, to the best of our knowledge, these are first results in the literature with explicit error bounds. Moreover, we also discuss neural network architecture that can be suitable for approximating symmetric and anti-symmetric functions.}, - archiveprefix = {arXiv}, - arxivid = {1912.01765v1}, - author = {Han, J and Li, Y and Lin, L and Lu, J and Zhang, J and Zhang, L}, - date = {2019}, - eprint = {1912.01765v1}, - eprinttype = {arxiv}, - journaltitle = {arxiv.org}, - title = {Universal Approximation of Symmetric and Anti-Symmetric Functions}, - url = {https://arxiv.org/abs/1912.01765} -} - -@article{harmonValidatingFirstprinciplesMolecular2020a, - abstract = {Metal oxide/water interfaces play a crucial role in many electrochemical and photocatalytic processes, such as photoelectrochemical water splitting, the creation of fuel from sunlight, and electrochemical CO2 reduction. First-principles electronic structure calculations can reveal unique insights into these processes, such as the role of the alignment of the oxide electronic energy levels with those of liquid water. An essential prerequisite for the success of such calculations is the ability to predict accurate structural models of these interfaces, which in turn requires careful experimental validation. Here we report a general, quantitative validation protocol for first-principles molecular dynamics simulations of oxide/aqueous interfaces. The approach makes direct comparisons of interfacial x-ray reflectivity (XR) signals from experimental measurements and those obtained from ab initio simulations with semilocal and van der Waals functionals. The protocol is demonstrated here for the case of the Al2O3(001)/water interface, one of the simplest oxide/water interfaces. We discuss the technical requirements needed for validation, including the choice of the density functional, the simulation cell size, and the optimal choice of the thermodynamic ensemble. Our results establish a general paradigm for the validation of structural models and interactions at solid/water interfaces derived from first-principles simulations. While there is qualitative agreement between the simulated structures and the experimental best-fit structure, direct comparisons of simulated and measured XR intensities show quantitative discrepancies that derive from both bulk regions (i.e., alumina and water) as well as the interfacial region, highlighting the need for accurate density functionals to properly describe interfacial interactions. Our results show that XR data are sensitive not only to the atomic structure (i.e., the atom locations) but also to the electron-density distributions in both the substrate and at the interface.}, - annotation = {WOS:000591531700008}, - author = {Harmon, Katherine J. and Letchworth-Weaver, Kendra and Gaiduk, Alex P. and Giberti, Federico and Gygi, Francois and Chan, Maria K. Y. and Fenter, Paul and Galli, Giulia}, - date = {2020-11-16}, - doi = {10.1103/PhysRevMaterials.4.113805}, - issn = {2475-9953}, - journaltitle = {Physical Review Materials}, - langid = {english}, - location = {{College Pk}}, - number = {11}, - pages = {113805}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Mater.}, - title = {Validating First-Principles Molecular Dynamics Calculations of Oxide/Water Interfaces with x-Ray Reflectivity Data}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {4} -} - -@article{hattrick-simpersOpenCombinatorialDiffraction2021, - abstract = {Modern machine learning and autonomous experimentation schemes in materials science rely on accurate analysis of the data ingested by these models. Unfortunately, accurate analysis of the underlying data can be difficult, even for domain experts, complicating the training of the models intended to drive experiments. This is especially true when the goal is to identify the presence of weak signatures in diffraction or spectroscopic datasets. In this work, we examine a set of as-obtained diffraction data that track the phase transition from monoclinic to tetragonal in a Nb-doped VO2 film as a function of temperature and dopant concentration. We then task a set of domain experts and a set of machine learning experts with identifying which phase is present in each diffraction pattern manually and algorithmically, respectively; in both cases, the labels can vary dramatically, especially at the phase boundaries. We use the mode of the labels and the Shannon entropy as a method to capture, preserve and propagate consensus labels and their variance. Further we use the expert labels as a benchmark and demonstrate the use of Shannon entropy weighted scoring to test the performance of machine learning generated labels. Finally, we propose a material data challenge centered around generating improved labeling algorithms. This real-world dataset curated with expert labels can act as test bed for new algorithms. The raw data, annotations and code used in this study are all available online at data.gov and the interested reader is encouraged to replicate and improve the existing models}, - author = {Hattrick-Simpers, Jason R. and DeCost, Brian and Kusne, A. Gilad and Joress, Howie and Wong-Ng, Winnie and Kaiser, Debra L. and Zakutayev, Andriy and Phillips, Caleb and Sun, Shijing and Thapa, Janak and Yu, Heshan and Takeuchi, Ichiro and Buonassisi, Tonio}, - date = {2021-06-01}, - doi = {10/gkhbw2}, - issn = {2193-9772}, - journaltitle = {Integrating Materials and Manufacturing Innovation}, - langid = {english}, - number = {2}, - pages = {311--318}, - shortjournal = {Integr Mater Manuf Innov}, - title = {An {{Open Combinatorial Diffraction Dataset Including Consensus Human}} and {{Machine Learning Labels}} with {{Quantified Uncertainty}} for {{Training New Machine Learning Models}}}, - url = {https://doi.org/10.1007/s40192-021-00213-8}, - urldate = {2021-08-10}, - volume = {10} -} - -@article{hernandezFastAccurateTransferable2019a, - abstract = {The length and time scales of atomistic simulations are limited by the computational cost of the methods used to predict material properties. In recent years there has been great progress in the use of machine-learning algorithms to develop fast and accurate interatomic potential models, but it remains a challenge to develop models that generalize well and are fast enough to be used at extreme time and length scales. To address this challenge, we have developed a machine-learning algorithm based on symbolic regression in the form of genetic programming that is capable of discovering accurate, computationally efficient many-body potential models. The key to our approach is to explore a hypothesis space of models based on fundamental physical principles and select models within this hypothesis space based on their accuracy, speed, and simplicity. The focus on simplicity reduces the risk of overfitting the training data and increases the chances of discovering a model that generalizes well. Our algorithm was validated by rediscovering an exact Lennard-Jones potential and a Sutton-Chen embedded-atom method potential from training data generated using these models. By using training data generated from density functional theory calculations, we found potential models for elemental copper that are simple, as fast as embedded-atom models, and capable of accurately predicting properties outside of their training set. Our approach requires relatively small sets of training data, making it possible to generate training data using highly accurate methods at a reasonable computational cost. We present our approach, the forms of the discovered models, and assessments of their transferability, accuracy and speed.}, - annotation = {WOS:000496957400002}, - author = {Hernandez, Alberto and Balasubramanian, Adarsh and Yuan, Fenglin and Mason, Simon A. M. and Mueller, Tim}, - date = {2019-11-18}, - doi = {10.1038/s41524-019-0249-1}, - journaltitle = {Npj Computational Materials}, - langid = {english}, - location = {{London}}, - pages = {112}, - publisher = {{Nature Publishing Group}}, - shortjournal = {npj Comput. Mater.}, - title = {Fast, Accurate, and Transferable Many-Body Interatomic Potentials by Symbolic Regression}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {5} -} - -@article{herrCompressingPhysicalProperties2019, - abstract = {The answers to many unsolved problems lie in the intractable chemical space of molecules and materials. Machine learning techniques are rapidly growing in popularity as a way to compress and explore chemical space efficiently. One of the most important aspects of machine learning techniques is representation through the feature vector, which should contain the most important descriptors necessary to make accurate predictions, not least of which is the atomic species in the molecule or material. In this work we introduce a compressed representation of physical properties for atomic species we call the elemental modes. The elemental modes provide an excellent representation by capturing many of the nuances of the periodic table and the similarity of atomic species. We apply the elemental modes to several different tasks for machine learning algorithms and show that they enable us to make improvements to these tasks even beyond simply achieving higher accuracy predictions.}, - archiveprefix = {arXiv}, - author = {Herr, John E. and Koh, Kevin and Yao, Kun and Parkhill, John}, - date = {2019-08-28}, - doi = {10/ggb5bq}, - eprint = {1811.00123}, - eprinttype = {arxiv}, - issn = {0021-9606, 1089-7690}, - journaltitle = {The Journal of Chemical Physics}, - langid = {english}, - note = {Comment: 6 pages, 5 figures}, - number = {8}, - pages = {084103}, - shortjournal = {J. Chem. Phys.}, - title = {Compressing Physical Properties of Atomic Species for Improving Predictive Chemistry}, - url = {http://arxiv.org/abs/1811.00123}, - urldate = {2021-08-11}, - volume = {151} -} - -@article{herrCompressingPhysicsAutoencoder2019a, - abstract = {We define a vector quantity which corresponds to atomic species identity by compressing a set of physical properties with an autoencoder. This vector, referred to here as the elemental modes, provides many advantages in downstream machine learning tasks. Using the elemental modes directly as the feature vector, we trained a neural network to predict formation energies of elpasolites with improved accuracy over previous works on the same task. Combining the elemental modes with geometric features used in high-dimensional neural network potentials (HD-NNPs) solves many problems of scaling and efficiency in the development of such neural network potentials. Whereas similar models in the past have been limited to typically four atomic species (H, C, N, and O), our implementation does not scale in cost by adding more atomic species and allows us to train an HD-NNP model which treats molecules containing H, C, N, O, F, P, S, Cl, Se, Br, and I. Finally, we establish that our implementation allows us to define feature vectors for alchemical intermediate states in the HD-NNP model, which opens up new possibilities for performing alchemical free energy calculations on systems where bond breaking/forming is important. Published under license by AIP Publishing.}, - annotation = {WOS:000483889300007}, - author = {Herr, John E. and Koh, Kevin and Yao, Kun and Parkhill, John}, - date = {2019-08-28}, - doi = {10.1063/1.5108803}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {8}, - pages = {084103}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - shorttitle = {Compressing Physics with an Autoencoder}, - title = {Compressing Physics with an Autoencoder: {{Creating}} an Atomic Species Representation to Improve Machine Learning Models in the Chemical Sciences}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {151} -} - -@article{Hodapp2020, - abstract = {and a.shapeev@skoltech.ru Abstract A well-known drawback of state-of-the-art machine-learning interatomic potentials is their poor ability to extrapolate beyond the training domain. For small-scale problems with tens to hundreds of atoms this can be solved by using active learning which is able to select atomic configurations on which a potential attempts extrapolation and add them to the ab initio-computed training set. In this sense an active learning algorithm can be viewed as an on-the-fly interpolation of an ab initio model. For large-scale problems, possibly involving tens of thousands of atoms, this is not feasible because one cannot afford even a single density functional theory (DFT) computation with such a large number of atoms. This work marks a new milestone toward fully automatic ab initio-accurate large-scale atomistic simulations. We develop an active learning algorithm that identifies local subregions of the simulation region where the potential extrapolates. Then the algorithm constructs periodic configurations out of these local, non-periodic subregions, sufficiently small to be computable with plane-wave DFT codes, in order to obtain accurate ab initio energies. We benchmark our algorithm on the problem of screw dislocation motion in bcc tungsten and show that our algorithm reaches ab initio accuracy, down to typical magnitudes of numerical noise in DFT codes. We show that our algorithm reproduces material properties such as core structure, Peierls barrier, and Peierls stress. This unleashes new capabilities for computational materials science toward applications which have currently been out of scope if approached solely by ab initio methods.}, - author = {Hodapp, M and And, A Shapeev - Machine Learning: Science and 2020, undefined}, - date = {2020}, - doi = {10.1088/2632-2153/aba373}, - journaltitle = {iopscience.iop.org}, - title = {In Operando Active Learning of Interatomic Interaction during Large-Scale Simulations}, - url = {https://iopscience.iop.org/article/10.1088/2632-2153/aba373/meta} -} - -@article{hodappMachinelearningPotentialsEnable2021, - abstract = {We present an automated procedure for computing stacking fault energies in random alloys from large-scale simulations using moment tensor potentials (MTPs) with the accuracy of density functional theory (DFT). To that end, we develop an algorithm for training MTPs on random alloys. In the first step, our algorithm constructs a set of \textasciitilde 10000 or more training candidate configurations with 50-100 atoms that are representative for the atomic neighborhoods occurring in the large-scale simulation. In the second step, we use active learning to reduce this set to \textasciitilde 100 most distinct configurations - for which DFT energies and forces are computed and on which the potential is ultimately trained. We validate our algorithm for the MoNbTa medium-entropy alloy by showing that the MTP reproduces the DFT \$\textbackslash frac\{1\}\{4\}[111]\$ unstable stacking fault energy over the entire compositional space up to a few percent. Contrary to state-of-the-art methods, e.g., the coherent potential approximation (CPA) or special quasi-random structures (SQSs), our algorithm naturally accounts for relaxation, is not limited by DFT cell sizes, and opens opportunities to efficiently investigate follow-up problems, such as chemical ordering. In a broader sense, our algorithm can be easily modified to compute related properties of random alloys, for instance, misfit volumes, or grain boundary energies. Moreover, it forms the basis for an efficient construction of MTPs to be used in large-scale simulations of multicomponent systems.}, - archiveprefix = {arXiv}, - author = {Hodapp, Max and Shapeev, Alexander}, - date = {2021-07-12}, - eprint = {2107.05620}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: preprint, 15 pages}, - primaryclass = {cond-mat, physics:physics}, - title = {Machine-Learning Potentials Enable Predictive \$\textbackslash textit\{and\}\$ Tractable High-Throughput Screening of Random Alloys}, - url = {http://arxiv.org/abs/2107.05620}, - urldate = {2021-08-11} -} - -@article{houDielectricConstantSupercritical2020, - abstract = {A huge amount of water at supercritical conditions exists in Earth's interior, where its dielectric properties play a critical role in determining how it stores and transports materials. However, it is very challenging to obtain the static dielectric constant of water, E-0, in a wide pressure-temperature (P-T) range as found in deep Earth either experimentally or by first-principles simulations. Here, we introduce a neural network dipole model, which, combined with molecular dynamics, can be used to compute P-T dependent dielectric properties of water as accurately as first-principles methods but much more efficiently. We found that E-0 may vary by one order of magnitude in Earth's upper mantle, suggesting that the solvation properties of water change dramatically at different depths. Although E-0 and the molecular dipole moment increase with an increase in pressure along an isotherm, the dipolar angular correlation has its maximum at 5 GPa-7 GPa, which may indicate that hydrogen bonds become weaker at high pressure. We also calculated the frequency-dependent dielectric constant of water in the microwave range, which, to the best of our knowledge, has not been calculated from first principles, and found that temperature affects the dielectric absorption more than pressure. Our results are of great use in many areas, e.g., modeling water-rock interactions in geochemistry. The computational approach introduced here can be readily applied to other molecular fluids.}, - annotation = {WOS:000571925300001}, - author = {Hou, Rui and Quan, Yuhui and Pan, Ding}, - date = {2020-09-14}, - doi = {10.1063/5.0020811}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {10}, - pages = {101103}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Dielectric Constant of Supercritical Water in a Large Pressure-Temperature Range}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{huangDeepPotentialGeneration2021, - abstract = {Solid-state electrolyte materials with superior lithium ionic conductivities are vital to the next-generation Li-ion batteries. Molecular dynamics could provide atomic scale information to understand the diffusion process of Li-ion in these superionic conductor materials. Here, we implement the deep potential generator to set up an efficient protocol to automatically generate interatomic potentials for Li10GeP2S12-type solid-state electrolyte materials (Li10GeP2S12, Li10SiP2S12, and Li10SnP2S12). The reliability and accuracy of the fast interatomic potentials are validated. With the potentials, we extend the simulation of the diffusion process to a wide temperature range (300 K-1000 K) and systems with large size (similar to 1000 atoms). Important technical aspects such as the statistical error and size effect are carefully investigated, and benchmark tests including the effect of density functional, thermal expansion, and configurational disorder are performed. The computed data that consider these factors agree well with the experimental results, and we find that the three structures show different behaviors with respect to configurational disorder. Our work paves the way for further research on computation screening of solid-state electrolyte materials.}, - annotation = {WOS:000630524800003}, - author = {Huang, Jianxing and Zhang, Linfeng and Wang, Han and Zhao, Jinbao and Cheng, Jun and Weinan, E.}, - date = {2021-03-07}, - doi = {10.1063/5.0041849}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {9}, - pages = {094703}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Deep Potential Generation Scheme and Simulation Protocol for the {{Li10GeP2S12}}-Type Superionic Conductors}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {154} -} - -@article{huangInitioMachineLearning2021, - abstract = {Chemical compound space (CCS), the set of all theoretically conceivable combinations of chemical elements and (meta-)stable geometries that make up matter, is colossal. The first principles based virtual sampling of this space, for example in search of novel molecules or materials which exhibit desirable properties, is therefore prohibitive for all but the smallest sub-sets and simplest properties. We review studies aimed at tackling this challenge using modern machine learning techniques based on (i) synthetic data, typically generated using quantum mechanics based methods, and (ii) model architectures inspired by quantum mechanics. Such Quantum mechanics based Machine Learning (QML) approaches combine the numerical efficiency of statistical surrogate models with an \{\textbackslash em ab initio\} view on matter. They rigorously reflect the underlying physics in order to reach universality and transferability across CCS. While state-of-the-art approximations to quantum problems impose severe computational bottlenecks, recent QML based developments indicate the possibility of substantial acceleration without sacrificing the predictive power of quantum mechanics.}, - archiveprefix = {arXiv}, - author = {Huang, Bing and von Lilienfeld, O. Anatole}, - date = {2021-06-22}, - eprint = {2012.07502}, - eprinttype = {arxiv}, - langid = {english}, - options = {useprefix=true}, - primaryclass = {physics}, - title = {Ab Initio Machine Learning in Chemical Compound Space}, - url = {http://arxiv.org/abs/2012.07502}, - urldate = {2021-08-11} -} - -@article{huangIntDeepDeepLearning2020, - abstract = {This paper proposes a deep-learning-initialized iterative method (Int-Deep) for low-dimensional nonlinear partial differential equations (PDEs). The corresponding framework consists of two phases. In the first phase, an expectation minimization problem formulated from a given nonlinear PDE is approximately resolved with mesh-free deep neural networks to parametrize the solution space. In the second phase, a solution ansatz of the finite element method to solve the given PDE is obtained from the approximate solution in the first phase, and the ansatz can serve as a good initial guess such that Newton’s method or other iterative methods for solving the nonlinear PDE are able to converge to the ground truth solution with high-accuracy quickly. Systematic theoretical analysis is provided to justify the Int-Deep framework for several classes of problems. Numerical results show that the Int-Deep outperforms existing purely deep learning-based methods or traditional iterative methods (e.g., Newton’s method and the Picard iteration method).}, - archiveprefix = {arXiv}, - author = {Huang, Jianguo and Wang, Haoqin and Yang, Haizhao}, - date = {2020-10}, - doi = {10/gg2rtj}, - eprint = {1910.01594}, - eprinttype = {arxiv}, - issn = {00219991}, - journaltitle = {Journal of Computational Physics}, - langid = {english}, - pages = {109675}, - shortjournal = {Journal of Computational Physics}, - shorttitle = {Int-{{Deep}}}, - title = {Int-{{Deep}}: {{A Deep Learning Initialized Iterative Method}} for {{Nonlinear Problems}}}, - url = {http://arxiv.org/abs/1910.01594}, - urldate = {2021-08-11}, - volume = {419} -} - -@article{huangLearningThermodynamicallyStable2021, - abstract = {In this work, we develop a method for learning interpretable, thermodynamically stable and Galilean invariant partial differential equations (PDEs) based on the conservation-dissipation formalism of irreversible thermodynamics. As governing equations for non-equilibrium flows in one dimension, the learned PDEs are parameterized by fully connected neural networks and satisfy the conservation-dissipation principle automatically. In particular, they are hyperbolic balance laws and Galilean invariant. The training data are generated from a kinetic model with smooth initial data. Numerical results indicate that the learned PDEs can achieve good accuracy in a wide range of Knudsen numbers. Remarkably, the learned dynamics can give satisfactory results with randomly sampled discontinuous initial data and Sod's shock tube problem although it is trained only with smooth initial data.}, - author = {Huang, Juntao and Ma, Zhiting and Zhou, Yizhou and Yong, Wen An}, - date = {2021}, - doi = {10.1515/JNET-2021-0008/HTML}, - journaltitle = {Journal of Non-Equilibrium Thermodynamics}, - publisher = {{De Gruyter Open Ltd}}, - title = {Learning Thermodynamically Stable and Galilean Invariant Partial Differential Equations for Non-Equilibrium Flows}, - url = {https://www.degruyter.com/document/doi/10.1515/jnet-2021-0008/html} -} - -@article{huangMachineLearningMoment2021, - abstract = {In this paper, we take a data-driven approach and apply machine learning to the moment closure problem for radiative transfer equation in slab geometry. Instead of learning the unclosed high order moment, we propose to directly learn the gradient of the high order moment using neural networks. This new approach is consistent with the exact closure we derive for the free streaming limit and also provides a natural output normalization. A variety of benchmark tests, including the variable scattering problem, the Gaussian source problem and the two material problem, show both good accuracy and generalizability of our machine learning closure model.}, - archiveprefix = {arXiv}, - author = {Huang, Juntao and Cheng, Yingda and Christlieb, Andrew J. and Roberts, Luke F.}, - date = {2021-05-12}, - eprint = {2105.05690}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {physics}, - shorttitle = {Machine Learning Moment Closure Models for the Radiative Transfer Equation {{I}}}, - title = {Machine Learning Moment Closure Models for the Radiative Transfer Equation {{I}}: Directly Learning a Gradient Based Closure}, - url = {http://arxiv.org/abs/2105.05690}, - urldate = {2021-08-11} -} - -@article{huInclusionMachineLearning2018, - abstract = {We discuss a theoretical approach that employs machine learning potential energy surfaces (ML-PESs) in the nonadiabatic dynamics simulation of polyatomic systems by taking 6-aminopyrimidine as a typical example. The Zhu-Nakamura theory is employed in the surface hopping dynamics, which does not require the calculation of the nonadiabatic coupling vectors. The kernel ridge regression is used in the construction of the adiabatic PESs. In the nonadiabatic dynamics simulation, we use ML-PESs for most geometries and switch back to the electronic structure calculations for a few geometries either near the S-1/S-0 conical intersections or in the out-of-confidence regions. The dynamics results based on ML-PESs are consistent with those based on CASSCF PESs. The ML-PESs are further used to achieve the highly efficient massive dynamics simulations with a large number of trajectories. This work displays the powerful role of ML methods in the nonadiabatic dynamics simulation of polyatomic systems.}, - annotation = {WOS:000435026100002}, - author = {Hu, Deping and Xie, Yu and Li, Xusong and Li, Lingyue and Lan, Zhenggang}, - date = {2018-06-07}, - doi = {10.1021/acs.jpclett.8b00684}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {11}, - pages = {2725--2732}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - title = {Inclusion of {{Machine Learning Kernel Ridge Regression Potential Energy Surfaces}} in {{On}}-the-{{Fly Nonadiabatic Molecular Dynamics Simulation}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/6}, - urldate = {2021-08-06}, - volume = {9} -} - -@article{huNeuralNetworkForce2020a, - abstract = {A method using machine learning (ML) is proposed to describe metal growth for simulations, which retains the accuracy of ab initio density functional theory (DFT) and results in a thousands-fold reduction in the computational time. This method is based on atomic energy decomposition from DFT calculations. Compared with other ML methods, our energy decomposition approach can yield much more information with the same DFT calculations. This approach is employed for the amorphous sodium system, where only 1000 DFT molecular dynamics images are enough for training an accurate model. The DFT and neural network potential (NNP) are compared for the dynamics to show that similar structural properties are generated. Finally, metal growth experiments from liquid to solid in a small and larger system are carried out to demonstrate the ability of using NNP to simulate the real growth process.}, - annotation = {WOS:000515424300026}, - author = {Hu, Qin and Weng, Mouyi and Chen, Xin and Li, Shucheng and Pan, Feng and Wang, Lin-Wang}, - date = {2020-02-20}, - doi = {10.1021/acs.jpclett.9b03780}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {4}, - pages = {364--1369}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - title = {Neural {{Network Force Fields}} for {{Metal Growth Based}} on {{Energy Decompositions}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{huPerspectiveMultiscaleSimulation2021a, - abstract = {Phonon-mediated thermal transport is inherently multi-scale. The wave-length of phonons (considering phonons as waves) is typically at the nanometer scale; the typical size of a phonon wave energy packet is tens of nanometers, while the phonon mean free path (MFP) can be as long as microns. At different length scales, the phonons will interact with structures of different feature sizes, which can be as small as 0D defects (point defects), short to medium range linear defects (dislocations), medium to large range 2D planar defects (stacking faults and twin boundaries), and large scale 3D defects (voids, inclusions, and various microstructures). The nature of multi-scale thermal transport is that there are different heat transfer physics across different length scales and in the meantime the physics crossing the different scales is interdependent and coupled. Since phonon behavior is usually mode dependent, thermal transport in materials with a combined micro-/nano-structure complexity becomes complicated, making modeling this kind of transport process very challenging. In this perspective, we first summarize the advantages and disadvantages of computational methods for mono-scale heat transfer and the state-of-the-art multi-scale thermal transport modeling. We then discuss a few important aspects of the future development of multi-scale modeling, in particular with the aid of modern machine learning and uncertainty quantification techniques. As more sophisticated theoretical and computational methods continue to advance thermal transport predictions, novel heat transfer physics and thermally functional materials will be discovered for the pertaining energy systems and technologies.}, - annotation = {WOS:000612961700001}, - author = {Hu, Ming and Yang, Zhonghua}, - date = {2021-01-21}, - doi = {10.1039/d0cp03372c}, - issn = {1463-9076}, - journaltitle = {Physical Chemistry Chemical Physics}, - langid = {english}, - location = {{Cambridge}}, - number = {3}, - pages = {1785--1801}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Phys. Chem. Chem. Phys.}, - title = {Perspective on Multi-Scale Simulation of Thermal Transport in Solids and Interfaces}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {23} -} - -@article{husicCoarseGrainingMolecular2020, - abstract = {Coarse graining enables the investigation of molecular dynamics for larger systems and at longer timescales than is possible at an atomic resolution. However, a coarse graining model must be formulated such that the conclusions we draw from it are consistent with the conclusions we would draw from a model at a finer level of detail. It has been proved that a force matching scheme defines a thermodynamically consistent coarse-grained model for an atomistic system in the variational limit. Wang et al. [ACS Cent. Sci. 5, 755 (2019)] demonstrated that the existence of such a variational limit enables the use of a supervised machine learning framework to generate a coarse-grained force field, which can then be used for simulation in the coarse-grained space. Their framework, however, requires the manual input of molecular features to machine learn the force field. In the present contribution, we build upon the advance of Wang et al. and introduce a hybrid architecture for the machine learning of coarse-grained force fields that learn their own features via a subnetwork that leverages continuous filter convolutions on a graph neural network architecture. We demonstrate that this framework succeeds at reproducing the thermodynamics for small biomolecular systems. Since the learned molecular representations are inherently transferable, the architecture presented here sets the stage for the development of machine-learned, coarse-grained force fields that are transferable across molecular systems.}, - annotation = {WOS:000595274800001}, - author = {Husic, Brooke E. and Charron, Nicholas E. and Lemm, Dominik and Wang, Jiang and Perez, Adria and Majewski, Maciej and Kramer, Andreas and Chen, Yaoyi and Olsson, Simon and de Fabritiis, Gianni and Noe, Frank and Clementi, Cecilia}, - date = {2020-11-21}, - doi = {10.1063/5.0026133}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {19}, - options = {useprefix=true}, - pages = {194101}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Coarse Graining Molecular Dynamics with Graph Neural Networks}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{ideensadrehaghighiArtificialNeutralNetworks2021, - abstract = {Evolution strategies [Rechenberg] such as ANNs, are optimization techniques that avoid the problems associated with the use of gradients as they require only the calculation of the cost function at each point in the parameter space. They operate based on natural principles of evolution such as mutation, recombination, and selection. These operations are adapted so that the algorithm automatically develops and attempts to optimize a model landscape relating the cost function to its parameters. Compared with gradient based techniques, their convergence rate is usually much lower, thus requiring large numbers of iterations that could be unrealistic for some problems of engineering interest. On the other hand, they are highly parallel algorithms that efficiently exploit today's powerful parallel computer architectures and they are more likely than gradient based algorithms to identify a global optimum. This latter aspect makes them attractive in many engineering applications where the fitness landscape cannot be assumed uni-modal.}, - author = {Ideen Sadrehaghighi}, - date = {2021}, - doi = {10/gmf5vh}, - langid = {english}, - publisher = {{Unpublished}}, - title = {Artificial {{Neutral Networks}} ({{ANNs}}) {{Applied}} as {{CFD Optimization Techniques}}}, - url = {http://rgdoi.net/10.13140/RG.2.2.21827.14885/2}, - urldate = {2021-08-10}, - version = {2} -} - -@article{jacksonEfficientMultiscaleOptoelectronic2020, - abstract = {Conjugated polymers represent a high-potential material class due to the tunability of their optoelectronic properties via straightforward processing protocols. However, to in silico tailor these optoelectronic properties, one must compute the conformationally dependent electronic structure over mesoscopic length and time scales. This task represents a challenging multiscale computational problem in which quantum chemistry and atomistic or coarse-grained molecular dynamics must be integrated. Recently, we introduced a methodology, called electronic coarse graining (ECG), that utilizes artificial neural networks to compute ab initio quality electronic structures using only the reduced degrees of freedom of coarse-grained models. Here, we adapt ECG to variable-molecular weight conjugated oligomers by casting the problem in the framework of sequence detection. A machine-learning (ML) architecture employing one-dimensional convolutional neural networks and long short-term memory networks is utilized to compute ground-state orbital energies, charge density distributions, and optical spectra solely from the coarse-grained model's configurational degrees of freedom. Robust molecular weight transferability for ECG is established via a A-ML approach that leverages model electronic Hamiltonians for ground and excited-state property determination. The accuracy and transferability of the ECG methodology opens the door for scalable optoelectronic property prediction in conjugated polymers directly from coarse-grained degrees of freedom.}, - annotation = {WOS:000507721500051}, - author = {Jackson, Nicholas E. and Bowen, Alec S. and de Pablo, Juan J.}, - date = {2020-01-14}, - doi = {10.1021/acs.macromol.9b02020}, - issn = {0024-9297}, - journaltitle = {Macromolecules}, - langid = {english}, - location = {{Washington}}, - number = {1}, - options = {useprefix=true}, - pages = {482--490}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {Macromolecules}, - title = {Efficient {{Multiscale Optoelectronic Prediction}} for {{Conjugated Polymers}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {53} -} - -@article{jacksonElectronicStructureCoarsegrained2019, - abstract = {Computational studies aimed at understanding conformationally dependent electronic structure in soft materials require a combination of classical and quantum-mechanical simulations, for which the sampling of conformational space can be particularly demanding. Coarse-grained (CG) models provide a means of accessing relevant time scales, but CG configurations must be back-mapped into atomistic representations to perform quantum-chemical calculations, which is computationally intensive and inconsistent with the spatial resolution of the CG models. A machine learning approach, denoted as artificial neural network electronic coarse graining (ANN-ECG), is presented here in which the conformationally dependent electronic structure of a molecule is mapped directly to CG pseudo-atom configurations. By averaging over decimated degrees of freedom, ANN-ECG accelerates simulations by eliminating backmapping and repeated quantum-chemical calculations. The approach is accurate, consistent with the CG spatial resolution, and can be used to identify computationally optimal CG resolutions.}, - annotation = {WOS:000462564300049}, - author = {Jackson, Nicholas E. and Bowen, Alec S. and Antony, Lucas W. and Webb, Michael A. and Vishwanath, Venkatram and de Pablo, Juan J.}, - date = {2019-03}, - doi = {10.1126/sciadv.aav1190}, - issn = {2375-2548}, - journaltitle = {Science Advances}, - langid = {english}, - location = {{Washington}}, - number = {3}, - options = {useprefix=true}, - pages = {eaav1190}, - publisher = {{Amer Assoc Advancement Science}}, - shortjournal = {Sci. Adv.}, - title = {Electronic Structure at Coarse-Grained Resolutions from Supervised Machine Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {5} -} - -@article{jacksonRecentAdvancesMachine2019, - abstract = {The multiscale design of soft materials requires an ensemble of computational techniques spanning quantum-chemistry to molecular dynamics to continuum modeling. The recent emergence of machine-learning (ML) and modern optimization algorithms has accelerated material property prediction, as well as stimulated the development of hybrid ML/molecular modeling methodologies capable of providing physical insights unobtainable from purely physics-based modeling and intuition. Such hybrid techniques also have important ramifications for the ML-enhanced interpretation of results from simulations and experiments alike. Leveraging ML techniques for the design of chemical or morphological structures based on a target property or functionality represents an exciting goal for the general area of soft materials, including polymers, liquid crystals, colloids, or biomolecules, to name a few representative classes of systems. Here, we provide a perspective on recent work using ML techniques of relevance for the multiscale design of soft materials and outline potential future directions of interest to the soft materials community.}, - annotation = {WOS:000472549500014}, - author = {Jackson, Nicholas E. and Webb, Michael A. and de Pablo, Juan J.}, - date = {2019-03}, - doi = {10.1016/j.coche.2019.03.005}, - issn = {2211-3398}, - journaltitle = {Current Opinion in Chemical Engineering}, - langid = {english}, - location = {{Oxford}}, - options = {useprefix=true}, - pages = {106--114}, - publisher = {{Elsevier Sci Ltd}}, - shortjournal = {Curr. Opin. Chem. Eng.}, - title = {Recent Advances in Machine Learning towards Multiscale Soft Materials Design}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {23} -} - -@article{jainMachineLearningMetallurgy2021, - abstract = {High-strength metal alloys achieve their performance via careful control of the nucleation, growth, and kinetics of precipitation. Alloy mechanical properties are then controlled by atomic scale phenomena such as shearing of the precipitates by dislocations. Atomistic modeling to understand the operative mechanisms requires length and timescales far larger than those accessible by first-principles methods. Here, a family of Behler-Parinello neural network potentials (NNPs) for the Al-Mg-Si system is developed to enable quantitative studies of Al-6xxx alloys. The NNP is trained on metallurgically important quantities computed by first-principles density function theory (DFT) leading to high-fidelity predictions of intermetallic compounds, elastic constants, dilute solid-solution energetics, precipitate/matrix interfaces, Al stacking fault energies, antisite defect energies, and other quantities. The generalized stacking fault energy surfaces for the three prevalent β′′ precipitate compositions in peak-aged Al-6xxx are then computed with the NNP, and are validated by DFT computations at key points. A preliminary examination of early stage clustering kinetics and energetics in Al-6xxx is then made, showing the formation of low-energy Mg-Si structures and the trapping of vacancies in these clusters. The NNP thus shows significant transferability across structures, making it a powerful approach for chemically accurate simulations of metallurgical phenomena in Al-Mg-Si alloys.}, - author = {Jain, Abhinav C.P. and Marchand, Daniel and Glensk, Albert and Ceriotti, M. and Curtin, W. A.}, - date = {2021-05}, - doi = {10.1103/physrevmaterials.5.053805}, - journaltitle = {Physical Review Materials}, - number = {5}, - publisher = {{American Physical Society}}, - title = {Machine Learning for Metallurgy {{III}}: {{A}} Neural Network Potential for {{Al}}-{{Mg}}-{{Si}}}, - volume = {5} -} - -@article{janetQuantitativeUncertaintyMetric2019a, - abstract = {Machine learning (ML) models, such as artificial neural networks, have emerged as a complement to high-throughput screening, enabling characterization of new compounds in seconds instead of hours. The promise of ML models to enable large-scale chemical space exploration can only be realized if it is straightforward to identify when molecules and materials are outside the model's domain of applicability. Established uncertainty metrics for neural network models are either costly to obtain (e.g., ensemble models) or rely on feature engineering (e.g., feature space distances), and each has limitations in estimating prediction errors for chemical space exploration. We introduce the distance to available data in the latent space of a neural network ML model as a low-cost, quantitative uncertainty metric that works for both inorganic and organic chemistry. The calibrated performance of this approach exceeds widely used uncertainty metrics and is readily applied to models of increasing complexity at no additional cost. Tightening latent distance cutoffs systematically drives down predicted model errors below training errors, thus enabling predictive error control in chemical discovery or identification of useful data points for active learning.}, - annotation = {WOS:000482978700005}, - author = {Janet, Jon Paul and Duan, Chenru and Yang, Tzuhsiung and Nandy, Aditya and Kulik, Heather J.}, - date = {2019-09-14}, - doi = {10.1039/c9sc02298h}, - issn = {2041-6520}, - journaltitle = {Chemical Science}, - langid = {english}, - location = {{Cambridge}}, - number = {34}, - pages = {7913--7922}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Chem. Sci.}, - title = {A Quantitative Uncertainty Metric Controls Error in Neural Network-Driven Chemical Discovery}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {10} -} - -@report{janetUncertainTimesCall2019, - abstract = {Machine learning (ML) models, such as artificial neural networks, have emerged as a complement to high-throughput screening, enabling characterization of new compounds in seconds instead of hours. The promise of ML models to enable large-scale, chemical space exploration can only be realized if it is straightforward to identify when molecules and materials are outside the model’s domain of applicability. Established uncertainty metrics for neural network models are either costly to obtain (e.g., ensemble models) or rely on feature engineering (e.g., feature space distances), and each has limitations in estimating prediction errors for chemical space exploration. We introduce the distance to available data in the latent space of a neural network ML model as a low-cost, quantitative uncertainty metric that works for both inorganic and organic chemistry. The calibrated performance of this approach exceeds widely used uncertainty metrics and is readily applied to models of increasing complexity at no additional cost. Tightening latent distance cutoffs systematically drives down predicted model errors below training errors, thus enabling predictive error control in chemical discovery or identification of useful data points for active learning.}, - author = {Janet, Jon Paul and Duan, Chenru and Yang, Tzuhsiung and Nandy, Aditya and Kulik, Heather}, - date = {2019-03-27}, - doi = {10.26434/chemrxiv.7900277.v1}, - langid = {english}, - shorttitle = {Uncertain {{Times Call}} for {{Quantitative Uncertainty Metrics}}}, - title = {Uncertain {{Times Call}} for {{Quantitative Uncertainty Metrics}}: {{Controlling Error}} in {{Neural Network Predictions}} for {{Chemical Discovery}}}, - type = {preprint}, - url = {https://chemrxiv.org/articles/Uncertain_Times_Call_for_Quantitative_Uncertainty_Metrics_Controlling_Error_in_Neural_Network_Predictions_for_Chemical_Discovery/7900277/1}, - urldate = {2021-08-11} -} - -@article{Jiang, - abstract = {Atmospheric aerosol nucleation contributes to around half of cloud condensation nuclei (CCN) globally and the nucleated particles can grow larger to impact air quality and consequently human health. Despite the decades' efforts, the detailed nucleation mechanism is still poorly understood. The ultimate goal of theoretical understanding aerosol nucleation is to simulate nucleation in ambient condition. However, there is lack of accurate reactive force field. Here for the first time, we propose the reactive force field with good size scalability for nucleation systems based on deep neural network and further bridge the simulation in the limited box with cluster kinetics towards boosting the aerosol simulation to be fully ab initio. We found that the formation rates based on hard sphere collision rate constants tend to be underestimated. Besides, the framework here is transferable to other nucleation systems, potentially revolutionizing the atmospheric aerosol nucleation field. Theoretical understanding of nucleation mechanism largely relies on classical nucleation theory (CNT) 1 which give a general mind map for nucleation thermodynamics and kinetics 2 even though the capillary assumption is extensively criticized 3. The emergence and then broadly employed theoretical model in the past decade, Atmospheric Cluster Dynamics Code (ACDC) 4 , surmounts the drawbacks of CNT through coupled quantum chemical thermodynamics 5 with the birth-death equations 2. Within the model, collision rate constants and evaporation rates are the two most critical parameters, determining the accuracy to predict macro parameters like cluster concentrations and formation rates which can be directly compared with experiments 6. Evaporation rates, derived from detailed balance and ab initio thermodynamics can be very accurate with sophisticated theoretical calculations 7. However, collision rate constants, derived from simple hard sphere model, is still very rough, far from the ab initio based evaporation rate. Here we propose a general framework to potentially boost the aerosol nucleation simulation toward fully ab initio. In the framework, deep neural network based force field (DNN-FF) is trained that can perform robust nucleation molecular dynamics (MD) simulations to derive the collision rate constants. Then static quantum chemical thermodynamics based evaporation rates are coupled into cluster dynamics model to provide ab initio kinetics for atmospheric aerosol nucleation.}, - author = {Jiang, S and Liu, YR and Huang, T and Feng, YJ and Wang, CY}, - date = {2021}, - journaltitle = {arxiv.org}, - title = {Towards Fully Ab Initio Simulation of Atmospheric Aerosol Nucleation}, - url = {https://arxiv.org/abs/2107.04802} -} - -@article{jiangAccurateDeepPotential2021, - abstract = {Combining first-principles accuracy and empirical-potential efficiency for the description of the potential energy surface (PES) is the philosopher's stone for unraveling the nature of matter via atomistic simulation. This has been particularly challenging for multi-component alloy systems due to the complex and non-linear nature of the associated PES. In this work, we develop an accurate PES model for the Al-Cu-Mg system by employing Deep Potential (DP), a neural network based representation of the PES, and DP Generator (DP-GEN), a concurrent-learning scheme that generates a compact set of ab initio data for training. The resulting DP model gives predictions consistent with first-principles calculations for various binary and ternary systems on their fundamental energetic and mechanical properties, including formation energy, equilibrium volume, equation of state, interstitial energy, vacancy and surface formation energy, as well as elastic moduli. Extensive benchmark shows that the DP model is ready and will be useful for atomistic modeling of the Al-Cu-Mg system within the full range of concentration.}, - archiveprefix = {arXiv}, - arxivid = {2008.11795v2}, - author = {Jiang, W and Zhang, Y and Zhang, L and H, Wang}, - date = {2021}, - eprint = {2008.11795v2}, - eprinttype = {arxiv}, - isbn = {2008.11795v2}, - journaltitle = {iopscience.iop.org}, - title = {Accurate {{Deep Potential}} Model for the {{Al}}–{{Cu}}–{{Mg}} Alloy in the Full Concentration Space}, - url = {https://iopscience.iop.org/article/10.1088/1674-1056/abf134/meta} -} - -@article{jiangAccurateDeepPotential2021a, - abstract = {Combining first-principles accuracy and empirical-potential efficiency for the description of the potential energy surface (PES) is the philosopher's stone for unraveling the nature of matter via atomistic simulation. This has been particularly challenging for multi-component alloy systems due to the complex and non-linear nature of the associated PES. In this work, we develop an accurate PES model for the Al-Cu-Mg system by employing deep potential (DP), a neural network based representation of the PES, and DP generator (DP-GEN), a concurrent-learning scheme that generates a compact set of ab initio data for training. The resulting DP model gives predictions consistent with first-principles calculations for various binary and ternary systems on their fundamental energetic and mechanical properties, including formation energy, equilibrium volume, equation of state, interstitial energy, vacancy and surface formation energy, as well as elastic moduli. Extensive benchmark shows that the DP model is ready and will be useful for atomistic modeling of the Al-Cu-Mg system within the full range of concentration.}, - annotation = {WOS:000655293300001}, - author = {Jiang, Wanrun and Zhang, Yuzhi and Zhang, Linfeng and Wang, Han}, - date = {2021-05}, - doi = {10.1088/1674-1056/abf134}, - issn = {1674-1056}, - journaltitle = {Chinese Physics B}, - langid = {english}, - location = {{Bristol}}, - number = {5}, - pages = {050706}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Chin. Phys. B}, - title = {Accurate {{Deep Potential}} Model for the {{Al}}-{{Cu}}-{{Mg}} Alloy in the Full Concentration Space}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {30} -} - -@article{jiaoSelfhealingMechanismLithium2021, - abstract = {Li metal as an ideal anode material meets the requirements of state-of-the-art secondary batteries. However, the dendrite growth of Li causes safety concerns and results in a low coulombic efficiency, which has significantly restricted the commercial applications of Li secondary batteries. Owing to the intrinsic limitations of even the most advanced experimental and computational techniques, a mechanistic understanding of Li deposition (growth) on the atomic scale is lacking. Here, we construct a Li potential model by machine learning with an accuracy of quantum mechanical computations. Our molecular dynamics simulations based on this potential model reveal two self-healing mechanisms in a large Li metal system, surface self-healing and bulk self-healing, and identify three Li dendrite morphologies in different conditions, "needle," "mushroom," and "hemisphere." We finally propose the concepts of local current density and variance of local current density as a supplement to critical current density to determine the probability of self-healing triggered.}, - archiveprefix = {arXiv}, - author = {Jiao, Junyu and Lai, Genming and Lu, Jiaze and Xu, Xianqi and Wang, Jing and Zheng, Jiaxin}, - date = {2021-06-21}, - eprint = {2106.10979}, - eprinttype = {arxiv}, - primaryclass = {cond-mat, physics:physics}, - title = {Self-Healing Mechanism of Lithium Metal}, - url = {http://arxiv.org/abs/2106.10979}, - urldate = {2021-08-11} -} - -@article{jiaPushingLimitMolecular2021, - abstract = {For 35 years, ab initio molecular dynamics (AIMD) has been the method of choice for modeling complex atom-istic phenomena from first principles. However, most AIMD applications are limited by computational cost to systems with thousands of atoms at most. We report that a machine learning-based simulation protocol (Deep Potential Molecular Dynamics), while retaining ab initio accuracy, can simulate more than 1 nanosecond-long trajectory of over 100 million atoms per day, using a highly optimized code (GPU DeePMD-kit) on the Summit supercomputer. Our code can efficiently scale up to the entire Summit supercomputer, attaining 91 PFLOPS in double precision (45.5\% of the peak) and 162/275 PFLOPS in mixed-single/half precision. The great accomplishment of this work is that it opens the door to simulating unprecedented size and time scales with ab initio accuracy. It also poses new challenges to the next-generation supercomputer for a better integration of machine learning and physical modeling.}, - archiveprefix = {arXiv}, - arxivid = {2005.00223v3}, - author = {Jia, Weile and Wang, Han and Chen, Mohan and Lu, Denghui and Lin, L and Lin, Lin and Car, Roberto and Zhang, Linfeng}, - date = {2021}, - eprint = {2005.00223v3}, - eprinttype = {arxiv}, - isbn = {9781728199986}, - journaltitle = {ieeexplore.ieee.org}, - title = {Pushing the Limit of Molecular Dynamics with Ab Initio Accuracy to 100 Million Atoms with Machine Learning}, - url = {https://ieeexplore.ieee.org/abstract/document/9355242/} -} - -@article{jinnouchiOntheflyActiveLearning2020, - abstract = {The on-the-fly generation of machine-learning force fields by active-learning schemes attracts a great deal of attention in the community of atomistic simulations. The algorithms allow the machine to self-learn an interatomic potential and construct machine-learned models on the fly during simulations. State-of-the-art query strategies allow the machine to judge whether new structures are out of the training data set or not. Only when the machine judges the necessity of updating the data set with the new structures are first-principles calculations carried out. Otherwise, the yet available machine-learned model is used to update the atomic positions. In this manner, most of the first-principles calculations are bypassed during training, and overall, simulations are accelerated by several orders of magnitude while retaining almost first-principles accuracy. In this Perspective, after describing essential components of the active-learning algorithms, we demonstrate the power of the schemes by presenting recent applications.}, - author = {Jinnouchi, Ryosuke and Miwa, Kazutoshi and Karsai, Ferenc and Kresse, Georg and Asahi, Ryoji}, - date = {2020-09}, - doi = {10.1021/acs.jpclett.0c01061}, - journaltitle = {Journal of Physical Chemistry Letters}, - number = {17}, - pages = {6946--6955}, - publisher = {{American Chemical Society}}, - title = {On-the-Fly Active Learning of Interatomic Potentials for Large-Scale Atomistic Simulations}, - volume = {11} -} - -@article{kangLargeScaleAtomicSimulation2020a, - abstract = {Atomic simulations based on quantum mechanics (QM) calculations have entered into the tool box of chemists over the past few decades, facilitating an understanding of a wide range of chemistry problems, from structure characterization to reactivity determination. Due to the poor scaling and high computational cost intrinsic to QM calculations, one has to either sacrifice accuracy or time when performing large-scale atomic simulations. The battle to find a better compromise between accuracy and speed has been central to the development of new theoretical methods. The recent advances of machine-learning (ML)-based large-scale atomic simulations has shown great promise to the benefit of many branches of chemistry. Instead of solving the Schrodinger equation directly, ML-based simulations rely on a large data set of accurate potential energy surfaces (PESs) and complex numerical models to predict the total energy. These simulations feature both a high speed and a high accuracy for computing large systems. Due to the lack of a physical foundation in numerical models, ML models are often frustrated in their predictivity and robustness, which are key to applications. Focusing on these concerns, here we overview the recent advances in ML methodologies for atomic simulations on three key aspects. Namely, the generation of a representative data set, the extensity of ML models, and the continuity of data representation. While global optimization methods are the natural choice for building a representative data set, the stochastic surface walking method is shown to provide the desired PES sampling for both minima and transition regions on the PES. The current ML models generally utilize local geometrical descriptors as an input and consider the total energy as the sum of atomic energies. There are many flavors of data descriptors and ML models, but the applications for material and reaction predictions are still limited, not least because of the difficulty to train the associated vast global data sets. We show that our recently designed power-type structure descriptors together with a feed-forward neural network (NN) model are compatible with highly complex global PES data, which has led to a large family of global NN (G-NN) potentials. Two recent applications of G-NN potentials in material and reaction simulations are selected to illustrate how ML-based atomic simulations can help the discovery of new materials and reactions.}, - annotation = {WOS:000584415900009}, - author = {Kang, Pei-Lin and Shang, Cheng and Liu, Zhi-Pan}, - date = {2020-10-20}, - doi = {10.1021/acs.accounts.0c00472}, - issn = {0001-4842}, - journaltitle = {Accounts of Chemical Research}, - langid = {english}, - location = {{Washington}}, - number = {10}, - pages = {2119--2129}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {Accounts Chem. Res.}, - title = {Large-{{Scale Atomic Simulation}} via {{Machine Learning Potentials Constructed}} by {{Global Potential Energy Surface Exploration}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {53} -} - -@article{kasamatsuEnablingInitioConfigurational2020, - abstract = {We propose a scheme for ab initio configurational sampling in multicomponent crystalline solids using Behler-Parinello type neural network potentials (NNPs) in an unconventional way: the NNPs are trained to predict the energies of relaxed structures from the perfect lattice with configurational disorder instead of the usual way of training to predict energies as functions of continuous atom coordinates. Training set bias is avoided through an active learning scheme. This idea is demonstrated on the calculation of the temperature dependence of the degree of A/B site inversion in MgAl\$\_2\$O\$\_4\$, which is a multivalent system requiring careful handling of long-range interactions. The present scheme may serve as an alternative to cluster expansion for `difficult' systems, e.g., complex bulk or interface systems with many components and sublattices that are relevant to many technological applications today.}, - author = {Kasamatsu, Shusuke and Motoyama, Yuichi and Yoshimi, Kazuyoshi and Matsumoto, Ushio and Kuwabara, Akihide and Ogawa, Takafumi}, - date = {2020-08-06}, - langid = {english}, - title = {Enabling Ab Initio Configurational Sampling of Multicomponent Solids with Long-Range Interactions Using Neural Network Potentials and Active Learning}, - url = {https://arxiv.org/abs/2008.02572v1}, - urldate = {2021-08-11} -} - -@article{keithCombiningMachineLearning2021, - abstract = {Machine learning models are poised to make a transformative impact on chemical sciences by dramatically accelerating computational algorithms and amplifying insights available from computational chemistry methods. However, achieving this requires a confluence and coaction of expertise in computer science and physical sciences. This review is written for new and experienced researchers working at the intersection of both fields. We first provide concise tutorials of computational chemistry and machine learning methods, showing how insights involving both can be achieved. We then follow with a critical review of noteworthy applications that demonstrate how computational chemistry and machine learning can be used together to provide insightful (and useful) predictions in molecular and materials modeling, retrosyntheses, catalysis, and drug design.}, - archiveprefix = {arXiv}, - author = {Keith, John A. and Vassilev-Galindo, Valentin and Cheng, Bingqing and Chmiela, Stefan and Gastegger, Michael and Müller, Klaus-Robert and Tkatchenko, Alexandre}, - date = {2021-02-11}, - eprint = {2102.06321}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 169 pages, 24 figures}, - primaryclass = {physics}, - title = {Combining {{Machine Learning}} and {{Computational Chemistry}} for {{Predictive Insights Into Chemical Systems}}}, - url = {http://arxiv.org/abs/2102.06321}, - urldate = {2021-08-11} -} - -@article{kimReactionPathForceMatching2021, - abstract = {First-principles determination of free energy profiles for condensed-phase chemical reactions is hampered by the daunting costs associated with configurational sampling on ab initio quantum mechanical/molecular mechanical (AI/MM) potential energy surfaces. Here, we report a new method that enables efficient AI/MM free energy simulations through mean force fitting. In this method, a free energy path in collective variables (CVs) is first determined on an efficient reactive aiding potential. Based on the configurations sampled along the free energy path, correcting forces to reproduce the AI/MM forces on the CVs are determined through force matching. The AI/MM free energy profile is then predicted from simulations on the aiding potential in conjunction with the correcting forces. Such cycles of correction–prediction are repeated until convergence is established. As the instantaneous forces on the CVs sampled in equilibrium ensembles along the free energy path are fitted, this procedure faithfully restores the target free energy profile by reproducing the free energy mean forces. Due to its close connection with the reaction path-force matching (RP-FM) framework recently introduced by us, we designate the new method as RP-FM in collective variables (RP-FM-CV). We demonstrate the effectiveness of this method on a type-II solution-phase SN2 reaction, NH3 + CH3Cl (the Menshutkin reaction), simulated with an explicit water solvent. To obtain the AI/MM free energy profiles, we employed the semiempirical AM1/MM Hamiltonian as the base level for determining the string minimum free energy pathway, along which the free energy mean forces are fitted to various target AI/MM levels using the Hartree–Fock (HF) theory, density functional theory (DFT), and the second-order Møller–Plesset perturbation (MP2) theory as the AI method. The forces on the bond-breaking and bond-forming CVs at both the base and target levels are obtained by force transformation from Cartesian to redundant internal coordinates under the Wilson B-matrix formalism, where the linearized FM is facilitated by the use of spline functions. For the Menshutkin reaction tested, our FM treatment greatly reduces the deviations on the CV forces, originally in the range of 12–33 to ∼2 kcal/mol/Å. Comparisons with the experimental and benchmark AI/MM results, tests of the new method under a variety of simulation protocols, and analyses of the solute–solvent radial distribution functions suggest that RP-FM-CV can be used as an efficient, accurate, and robust method for simulating solution-phase chemical reactions.}, - author = {Kim, Bryant and Snyder, Ryan and Nagaraju, Mulpuri and Zhou, Yan and Ojeda-May, Pedro and Keeton, Seth and Hege, Mellisa and Shao, Yihan and Pu, Jingzhi}, - date = {2021-08-10}, - doi = {10/gmfw5p}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - number = {8}, - pages = {4961--4980}, - publisher = {{American Chemical Society}}, - shortjournal = {J. Chem. Theory Comput.}, - shorttitle = {Reaction {{Path}}-{{Force Matching}} in {{Collective Variables}}}, - title = {Reaction {{Path}}-{{Force Matching}} in {{Collective Variables}}: {{Determining Ab Initio QM}}/{{MM Free Energy Profiles}} by {{Fitting Mean Force}}}, - url = {https://doi.org/10.1021/acs.jctc.1c00245}, - urldate = {2021-08-11}, - volume = {17} -} - -@article{Kocer2021, - abstract = {In the past two decades, machine learning potentials (MLP) have reached a level of maturity that now enables applications to large-scale atomistic simulations of a wide range of systems in chemistry, physics and materials science. Different machine learning algorithms have been used with great success in the construction of these MLPs. In this review, we discuss an important group of MLPs relying on artificial neural networks to establish a mapping from the atomic structure to the potential energy. In spite of this common feature, there are important conceptual differences, which concern the dimensionality of the systems, the inclusion of long-range electrostatic interactions, global phenomena like non-local charge transfer, and the type of descriptor used to represent the atomic structure, which can either be predefined or learnable. A concise overview is given along with a discussion of the open challenges in the field.}, - author = {Kocer, Emir and Ko, TW Tsz Wai and Behler, Jörg and Behler, J}, - date = {2021-07}, - journaltitle = {arxiv.org}, - title = {Neural Network Potentials: {{A}} Concise Overview of Methods}, - url = {https://arxiv.org/abs/2107.03727 http://arxiv.org/abs/2107.03727} -} - -@article{koEnablingLargeScaleCondensedPhase2020a, - abstract = {By including a fraction of exact exchange (EXX), hybrid functionals reduce the self-interaction error in semilocal density functional theory (DFT) and thereby furnish a more accurate and reliable description of the underlying electronic structure in systems throughout biology, chemistry, physics, and materials science. However, the high computational cost associated with the evaluation of all required EXX quantities has limited the applicability of hybrid DFT in the treatment of large molecules and complex condensed-phase materials. To overcome this limitation, we describe a linear-scaling approach that utilizes a local representation of the occupied orbitals (e.g., maximally localized Wannier functions (MLWFs)) to exploit the sparsity in the real-space evaluation of the quantum mechanical exchange interaction in finite-gap systems. In this work, we present a detailed description of the theoretical and algorithmic advances required to perform MLWF-based ab initio molecular dynamics (AIMD) simulations of large-scale condensed-phase systems of interest at the hybrid DFT level. We focus our theoretical discussion on the integration of this approach into the framework of Car-Parrinello AIMD, and highlight the central role played by the MLWF-product potential (i.e., the solution of Poisson's equation for each corresponding MLWF-product density) in the evaluation of the EXX energy and wave function forces. We then provide a comprehensive description of the exx algorithm implemented in the open-source Quantum ESPRESSO program, which employs a hybrid MPI/OpenMP parallelization scheme to efficiently utilize the high-performance computing (HPC) resources available on current- and next-generation supercomputer architectures. This is followed by a critical assessment of the accuracy and parallel performance (e.g., strong and weak scaling) of this approach when AIMD simulations of liquid water are performed in the canonical (NVT) ensemble. With access to HPC resources, we demonstrate that exx enables hybrid DFT-based AIMD simulations of condensed-phase systems containing 500-1000 atoms (e.g., (H2O)(256)) with a wall time cost that is comparable to that of semilocal DFT. In doing so, exx takes us one step closer to routinely performing AIMD simulations of complex and large-scale condensed-phase systems for sufficiently long time scales at the hybrid DFT level of theory.}, - annotation = {WOS:000541503600027}, - author = {Ko, Hsin-Yu and Jia, Junteng and Santra, Biswajit and Wu, Xifan and Car, Roberto and DiStasio, Robert}, - date = {2020-06-09}, - doi = {10.1021/acs.jctc.9b01167}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {6}, - pages = {3757--3785}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {Enabling {{Large}}-{{Scale Condensed}}-{{Phase Hybrid Density Functional Theory Based Ab Initio Molecular Dynamics}}. 1. {{Theory}}, {{Algorithm}}, and {{Performance}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {16} -} - -@article{koIsotopeEffectsLiquid2019, - abstract = {A comprehensive microscopic understanding of ambient liquid water is a major challenge for ab initio simulations as it simultaneously requires an accurate quantum mechanical description of the underlying potential energy surface (PES) as well as extensive sampling of configuration space. Due to the presence of light atoms (e.g. or ), nuclear quantum fluctuations lead to observable changes in the structural properties of liquid water (e.g. isotope effects), and therefore provide yet another challenge for ab initio approaches. In this work, we demonstrate that the combination of dispersion-inclusive hybrid density functional theory (DFT), the Feynman discretised path-integral (PI) approach, and machine learning (ML) constitutes a versatile ab initio based framework that enables extensive sampling of both thermal and nuclear quantum fluctuations on a quite accurate underlying PES. In particular, we employ the recently developed deep potential molecular dynamics (DPMD) model ? a neural-network representation of the ab initio PES ? in conjunction with a PI approach based on the generalised Langevin equation (PIGLET) to investigate how isotope effects influence the structural properties of ambient liquid and . Through a detailed analysis of the interference differential cross sections as well as several radial and angular distribution functions, we demonstrate that this approach can furnish a semi-quantitative prediction of these subtle isotope effects. [GRAPHICS] .}, - annotation = {WOS:000490699600001}, - author = {Ko, Hsin-Yu and Zhang, Linfeng and Santra, Biswajit and Wang, Han and E, Weinan and DiStasio, Robert A. and Car, Roberto}, - date = {2019-11-17}, - doi = {10.1080/00268976.2019.1652366}, - issn = {0026-8976}, - journaltitle = {Molecular Physics}, - langid = {english}, - location = {{Abingdon}}, - number = {22}, - pages = {3269--3281}, - publisher = {{Taylor \& Francis Ltd}}, - shortjournal = {Mol. Phys.}, - title = {Isotope Effects in Liquid Water via Deep Potential Molecular Dynamics}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {117} -} - -@article{kondorNbodyNetworksCovariant2018, - abstract = {We describe N –body networks, a neural network architecture for learning the behavior and properties of complex many body physical systems. Our specific application is to learn atomic potential energy surfaces for use in molecular dynamics simulations. Our architecture is novel in that (a) it is based on a hierarchical decomposition of the many body system into subsytems (b) the activations of the network correspond to the internal state of each subsystem (c) the “neurons” in the network are constructed explicitly so as to guarantee that each of the activations is covariant to rotations (d) the neurons operate entirely in Fourier space, and the nonlinearities are realized by tensor products followed by Clebsch–Gordan decompositions. As part of the description of our network, we give a characterization of what way the weights of the network may interact with the activations so as to ensure that the covariance property is maintained.}, - archiveprefix = {arXiv}, - author = {Kondor, Risi}, - date = {2018-03-05}, - eprint = {1803.01588}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {cs}, - shorttitle = {N-Body {{Networks}}}, - title = {N-Body {{Networks}}: A {{Covariant Hierarchical Neural Network Architecture}} for {{Learning Atomic Potentials}}}, - url = {http://arxiv.org/abs/1803.01588}, - urldate = {2021-08-11} -} - -@article{kontolatiManifoldLearningCoarsegraining2021, - abstract = {We introduce a generalized machine learning framework to probabilistically parameterize upper-scale models in the form of nonlinear PDEs consistent with a continuum theory, based on coarse-grained atomistic simulation data of mechanical deformation and flow processes. The proposed framework utilizes a hypothesized coarse-graining methodology with manifold learning and surrogate-based optimization techniques. Coarsegrained high-dimensional data describing quantities of interest of the multiscale models are projected onto a nonlinear manifold whose geometric and topological structure is exploited for measuring behavioral discrepancies in the form of manifold distances. A surrogate model is constructed using Gaussian process regression to identify a mapping between stochastic parameters and distances. Derivative-free optimization is employed to adaptively identify a unique set of parameters of the upper-scale model capable of rapidly reproducing the system’s behavior while maintaining consistency with coarse-grained atomic-level simulations. The proposed method is applied to learn the parameters of the shear transformation zone (STZ) theory of plasticity that describes plastic deformation in amorphous solids as well as coarse-graining parameters needed to translate between atomistic and continuum representations. We show that the methodology is able to successfully link coarse-grained microscale simulations to macroscale observables and achieve a high-level of parity between the models across scales.}, - archiveprefix = {arXiv}, - author = {Kontolati, Katiana and Alix-Williams, Darius and Boffi, Nicholas M. and Falk, Michael L. and Rycroft, Chris H. and Shields, Michael D.}, - date = {2021-07-23}, - eprint = {2103.00779}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 34 pages, 12 figures, references added, Section 4 added, Section 2.1 updated}, - primaryclass = {physics}, - shorttitle = {Manifold Learning for Coarse-Graining Atomistic Simulations}, - title = {Manifold Learning for Coarse-Graining Atomistic Simulations: {{Application}} to Amorphous Solids}, - url = {http://arxiv.org/abs/2103.00779}, - urldate = {2021-08-11} -} - -@article{Korotaev2019, - abstract = {While lattice thermal conductivity is an important parameter for many technological applications, its calculation is a time-consuming task, especially for compounds with a complex crystal structure. In this paper, we solve this problem using machine learning interatomic potentials. These potentials trained on the density functional theory results and provide an accurate description of lattice dynamics. Additionally, active learning was applied to significantly reduce the number of expensive quantum-mechanical calculations required for training and increases reliability of the potential. The CoSb 3 skutterudite was considered as an example, and the solution of the Boltzmann transport equation for phonons was compared with the Green-Kubo method. We demonstrated that accurate and reliable potentials can be obtained by performing just a few hundred quantum-mechanical calculations. The potentials reproduce not only the vibrational spectrum, but also the lattice thermal conductivity, as calculated by various methods.}, - author = {Korotaev, P and Novoselov, I and Yanilkin, A and B, A Shapeev}, - date = {2019-10}, - doi = {10.1103/physrevb.100.144308}, - journaltitle = {APS}, - number = {14}, - pages = {144308}, - publisher = {{American Physical Society}}, - title = {Accessing Thermal Conductivity of Complex Compounds by Machine Learning Interatomic Potentials}, - url = {https://journals.aps.org/prb/abstract/10.1103/PhysRevB.100.144308}, - volume = {100} -} - -@article{krishnamoorthyDielectricConstantLiquid2021a, - abstract = {The static dielectric constant epsilon(0) and its temperature dependence for liquid water is investigated using neural network quantum molecular dynamics (NNQMD). We compute the exact dielectric constant in canonical ensemble from NNQMD trajectories using fluctuations in macroscopic polarization computed from maximally localized Wannier functions (MLWF). Two deep neural networks are constructed. The first, NNQMD, is trained on QMD configurations for liquid water under a variety of temperature and density conditions to learn potential energy surface and forces and then perform molecular dynamics simulations. The second network, NNMLWF, is trained to predict locations of MLWF of individual molecules using the atomic configurations from NNQMD. Training data for both the neural networks is produced using a highly accurate quantum-mechanical method, DFT-SCAN that yields an excellent description of liquid water. We produce 280 x 10(6) configurations of water at 7 temperatures using NNQMD and predict MLWF centers using NNMLWF to compute the polarization fluctuations. The length of trajectories needed for a converged value of the dielectric constant at 0 degrees C is found to be 20 ns (40 x 10(6) configurations with 0.5 fs time step). The computed dielectric constants for 0, 15, 30, 45, 60, 75, and 90 degrees C are in good agreement with experiments. Our scalable scheme to compute dielectric constants with quantum accuracy is also applicable to other polar molecular liquids.}, - annotation = {WOS:000655929800008}, - author = {Krishnamoorthy, Aravind and Nomura, Ken-ichi and Baradwaj, Nitish and Shimamura, Kohei and Rajak, Pankaj and Mishra, Ankit and Fukushima, Shogo and Shimojo, Fuyuki and Kalia, Rajiv and Nakano, Aiichiro and Vashishta, Priya}, - date = {2021-05-25}, - doi = {10.1103/PhysRevLett.126.216403}, - issn = {0031-9007}, - journaltitle = {Physical Review Letters}, - langid = {english}, - location = {{College Pk}}, - number = {21}, - pages = {216403}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Lett.}, - title = {Dielectric {{Constant}} of {{Liquid Water Determined}} with {{Neural Network Quantum Molecular Dynamics}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {126} -} - -@article{kuritzSizeTemperatureTransferability2018, - abstract = {A direct and local deep learning (DL) model for atomic forces is presented. We demonstrate the model performance in bulk aluminum, sodium, and silicon; and show that its errors are comparable to those found in state-of-the-art machine learning and DL models. We then analyze the model’s performance as a function of the number of neighbors included and show that one can ascertain physical attributes of the system from the analysis of the deep learning model’s behavior. Finally, we test the size scaling performance of the model, and the transferability between different temperatures, and show that our model performs well in both scaling to larger systems and high-to-low temperature predictability.}, - archiveprefix = {arXiv}, - author = {Kuritz, Natalia and Gordon, Goren and Natan, Amir}, - date = {2018-09-20}, - doi = {10/gkv2j9}, - eprint = {1804.01151}, - eprinttype = {arxiv}, - issn = {2469-9950, 2469-9969}, - journaltitle = {Physical Review B}, - langid = {english}, - number = {9}, - pages = {094109}, - shortjournal = {Phys. Rev. B}, - title = {Size and {{Temperature Transferability}} of {{Direct}} and {{Local Deep Neural Networks}} for {{Atomic Forces}}}, - url = {http://arxiv.org/abs/1804.01151}, - urldate = {2021-08-11}, - volume = {98} -} - -@article{kwonEstimationSecondVirial2021, - abstract = {We test the accuracy of the neural network interaction potentials against accurate thermodynamic data of He and N2. We perform single point energy calculations at CCSD(T) level, construct the intermolecular interactions by using neural networks, and perform Monte Carlo simulations to calculate virial coefficients. We find that the deviation of virial coefficients from previous studies is small for He while the difference is about 3~cm3mol−1 for N2. Our results show that when trying to obtain accurate thermodynamic data from the neural network interaction potentials, not only the level of quantum chemical calculations should be an important factor, but also at least thousands of training data points would be required.}, - author = {Kwon, Taejin and Song, Han Wook and Woo, Sam Yong and Kim, Jong-Ho and Sung, Bong June}, - date = {2021-08-01}, - doi = {10/gmf6ws}, - issn = {0301-0104}, - journaltitle = {Chemical Physics}, - langid = {english}, - pages = {111231}, - shortjournal = {Chemical Physics}, - title = {The Estimation of the Second Virial Coefficients of {{He}} and {{N2}} Based on Neural Network Potentials with Quantum Mechanical Calculations}, - url = {https://www.sciencedirect.com/science/article/pii/S0301010421001427}, - urldate = {2021-08-10}, - volume = {548} -} - -@article{leiMachinelearningbasedNonNewtonianFluid2020, - abstract = {We introduce a machine-learning-based framework for constructing continuum a non-Newtonian fluid dynamics model directly from a microscale description. Dumbbell polymer solutions are used as examples to demonstrate the essential ideas. To faithfully retain molecular fidelity, we establish a micro-macro correspondence via a set of encoders for the microscale polymer configurations and their macroscale counterparts, a set of nonlinear conformation tensors. The dynamics of these conformation tensors can be derived from the microscale model, and the relevant terms can be parametrized using machine learning. The final model, named the deep non-Newtonian model (DeePN2), takes the form of conventional non-Newtonian fluid dynamics models, with a generalized form of the objective tensor derivative that retains the microscale interpretations. Both the formulation of the dynamic equation and the neural network representation rigorously preserve the rotational invariance, which ensures the admissibility of the constructed model. Numerical results demonstrate the accuracy of DeePN2 where models based on empirical closures show limitations.}, - author = {Lei, Huan and Wu, Lei and Weinan, Weinan}, - date = {2020-10}, - doi = {10.1103/physreve.102.043309}, - journaltitle = {Physical Review E}, - number = {4}, - publisher = {{American Physical Society}}, - title = {Machine-Learning-Based Non-{{Newtonian}} Fluid Model with Molecular Fidelity}, - volume = {102} -} - -@article{leModelingElectrochemicalInterfaces2020a, - abstract = {The origin of the potential difference between the potential of zero charge of a metal/water interface and the work function of the metal is a recurring issue because it is related to how water interacts with metal surface in the absence of surface charge. Recently ab initio molecular dynamics method has been used to model electrochemical interfaces to study interfacial potential and the structure of interface water. Here, we will first introduce the computational standard hydrogen electrode method, which allows for ab initio determination of electrode potentials that can be directly compared with experiment. Then, we will review the recent progress from ab initio molecular dynamics simulation in understanding the interaction between water and metal and its impact on interfacial potential. Finally, we will give our perspective for future development of ab initio computational electrochemistry.}, - annotation = {WOS:000519095200020}, - author = {Le, Jia-Bo and Cheng, Jun}, - date = {2020-02}, - doi = {10.1016/j.coelec.2019.11.008}, - issn = {2451-9103}, - journaltitle = {Current Opinion in Electrochemistry}, - langid = {english}, - location = {{Amsterdam}}, - pages = {129--136}, - publisher = {{Elsevier}}, - shortjournal = {Curr. Opin. Electrochem.}, - shorttitle = {Modeling Electrochemical Interfaces from Ab Initio Molecular Dynamics}, - title = {Modeling Electrochemical Interfaces from Ab Initio Molecular Dynamics: Water Adsorption on Metal Surfaces at Potential of Zero Charge}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {19} -} - -@article{leoniNonclassicalNucleationPathways2021, - abstract = {The nucleation of crystals from the liquid melt is often characterized by a competition between different crystalline structures or polymorphs, and can result in nuclei with heterogeneous compositions. These mixed-phase nuclei can display nontrivial spatial arrangements, such as layered and onion-like structures, whose composition varies according to the radial distance, and which so far have been explained on the basis of bulk and surface free-energy differences between the competing phases. Here we extend the generality of these non-classical nucleation processes, showing that layered and onion-like structures can emerge solely based on structural fluctuations even in absence of free-energy differences. We consider two examples of competing crystalline structures, hcp and fcc forming in hard spheres, relevant for repulsive colloids and dense liquids, and the cubic and hexagonal diamond forming in water, relevant also for other group 14 elements such as carbon and silicon. We introduce a novel structural order parameter that combined with a neural network classification scheme allows us to study the properties of the growing nucleus from the early stages of nucleation. We find that small nuclei have distinct size fluctuations and compositions from the nuclei that emerge from the growth stage. The transition between these two regimes is characterized by the formation of onion-like structures, in which the composition changes with the distance from the center of the nucleus, similarly to what seen in two-step nucleation process.}, - archiveprefix = {arXiv}, - author = {Leoni, Fabio and Russo, John}, - date = {2021-05-12}, - eprint = {2105.05506}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {cond-mat}, - title = {Non-Classical Nucleation Pathways in Stacking-Disordered Crystals}, - url = {http://arxiv.org/abs/2105.05506}, - urldate = {2021-08-11} -} - -@article{leoniNonclassicalNucleationPathways2021a, - abstract = {The nucleation of crystals from liquid melt is often characterized by a competition between different crystalline structures or polymorphs and can result in nuclei with heterogeneous compositions. These mixed-phase nuclei can display nontrivial spatial arrangements, such as layered and onionlike structures, whose composition varies according to the radial distance, and which so far have been explained on the basis of bulk and surface free-energy differences between the competing phases. Here we extend the generality of these nonclassical nucleation processes, showing that layered and onionlike structures can emerge solely based on structural fluctuations even in the absence of free-energy differences. We consider two examples of competing crystalline structures, hcp and fcc forming in hard spheres relevant for repulsive colloids and dense liquids, and the cubic and hexagonal diamond forming in water relevant also for other group 14 elements such as carbon and silicon. We introduce a novel structural order parameter that combined with a neural-network classification scheme allows us to study the properties of the growing nucleus from the early stages of nucleation. We find that small nuclei have distinct size fluctuations and compositions from the nuclei that emerge from the growth stage. The transition between these two regimes is characterized by the formation of onionlike structures, in which the composition changes with the distance from the center of the nucleus, similar to what is seen in the two-step nucleation process.}, - annotation = {WOS:000671589300001}, - author = {Leoni, Fabio and Russo, John}, - date = {2021-07-09}, - doi = {10.1103/PhysRevX.11.031006}, - issn = {2160-3308}, - journaltitle = {Physical Review X}, - langid = {english}, - location = {{College Pk}}, - number = {3}, - pages = {031006}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. X}, - title = {Nonclassical {{Nucleation Pathways}} in {{Stacking}}-{{Disordered Crystals}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{liAccurateTransferableReactive, - abstract = {Chemical reactions constitute the central feature of many liquid, material, and biomolecular processes. Conventional molecular dynamics (MD) is inadequate for simulating chemical reactions given the fixed bonding topology of most force fields, while modeling chemical reactions using ab initio molecular dynamics is limited to shorter time and length scales given its high computational cost. As such, the multiscale reactive molecular dynamics method provides one promising alternative for simulating complex chemical systems at atomistic detail on a reactive potential energy surface. However, the parameterization of such models is a key barrier to their applicability and success. In this work, we present reactive MD models derived from constrained density functional theory that are both accurate and transferable. We illustrate the features of these models for proton dissociation reactions of amino acids in both aqueous and protein environments. Specifically, we present models for ionizable glutamate and lysine that predict accurate absolute pKa values in water, as well as their significantly shifted pKa in staphylococcal nuclease (SNase) without any modification of the models. As one outcome of the new methodology, the simulations show that the deprotonation of ionizable residues in SNase can be closely coupled with sidechain rotations, which is a concept likely generalizable to many other proteins. Furthermore, the present approach is not limited to only pKa prediction, but can enable the fully atomistic simulation of many other reactive systems along with a determination of the key aspects of the reaction mechanisms.}, - author = {Li, Chenghan and Voth, Gregory A}, - langid = {english}, - pages = {31}, - title = {Accurate and {{Transferable Reactive Molecular Dynamics Models}} from {{Constrained Density Functional Theory}}} -} - -@article{liAnalysisTrajectorySimilarity2018, - abstract = {We propose an "automatic" approach to analyze the results of the on-the-fly trajectory surface hopping simulation on the multi-channel nonadiabatic photoisomerization dynamics by considering the trajectory similarity and the configuration similarity. We choose a representative system phytochromobilin (P Phi B) chromophore model to illustrate the analysis protocol. After a large number of trajectories are obtained, it is possible to define the similarity of different trajectories by the Frechet distance and to employ the trajectory clustering analysis to divide all trajectories into several clusters. Each cluster in principle represents a photoinduced isomerization reaction channel. This idea provides an effective approach to understand the branching ratio of the multi-channel photoisomerization dynamics. For each cluster, the dimensionality reduction is employed to understand the configuration similarity in the trajectory propagation, which provides the understanding of the major geometry evolution features in each reaction channel. The results show that this analysis protocol not only assigns all trajectories into different photoisomerization reaction channels but also extracts the major molecular motion without the requirement of the pre-known knowledge of the active photoisomerization site. As a side product of this analysis tool, it is also easy to find the so-called "typical" or "representative" trajectory for each reaction channel. Published by AIP Publishing.}, - annotation = {WOS:000454626000008}, - author = {Li, Xusong and Hu, Deping and Xie, Yu and Lan, Zhenggang}, - date = {2018-12-28}, - doi = {10.1063/1.5048049}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {24}, - pages = {244104}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Analysis of Trajectory Similarity and Configuration Similarity in On-the-Fly Surface-Hopping Simulation on Multi-Channel Nonadiabatic Photoisomerization Dynamics}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/6}, - urldate = {2021-08-06}, - volume = {149} -} - -@article{liangMachineLearningDrivenSimulationsMicrostructure2021, - abstract = {Theoretical studies on the MgCl2-KCl eutectic heavily rely on ab initio calculations based on density functional theory (DFT). However, neither large-scale nor long-time calculations are feasible in the framework of the ab initio method, which makes it challenging to accurately predict some properties. To address this issue, a scheme based on ab initio calculation, deep neural networks, and machine learning is introduced. By training on high-quality data sets generated by ab initio calculations, a deep potential (DP) is constructed to describe the interaction between atoms. This work shows that the DP enables higher efficiency and similar accuracy relative to DFT. By performing molecular dynamics simulations with DP, the microstructure and thermophysical properties of the MgCl2-KCl eutectic (32:68 mol \%) are investigated. The structural evolution with temperature is analyzed through partial radial distribution functions, coordination numbers, angular distribution functions, and structural factors. Meanwhile, the estimated thermophysical properties are discussed, including density, thermal expansion coefficient, shear viscosity, self-diffusion coefficient, and specific heat capacity. It reveals that the Mg2+ ions in this system have a distorted tetrahedral geometry rather than an octahedral one (with vacancies). The microstructure of the MgCl2-KCl eutectic shows the feature of medium-range order, and this feature will be enhanced at a higher temperature. All predicted thermophysical properties are in good agreement with the experimental results. The hydrodynamic radius determined from the shear viscosity and self-diffusion coefficient shows that the Me ions have a strong local structure and diffuse as if with an intact coordination shell. Overall, this work provides a thorough understanding of the microstructure and enriches the data of the thermophysical properties of the MgCl2-KCl eutectic.}, - annotation = {WOS:000614062400050}, - author = {Liang, Wenshuo and Lu, Guimin and Yu, Jianguo}, - date = {2021-01-27}, - doi = {10.1021/acsami.0c20665}, - issn = {1944-8244}, - journaltitle = {Acs Applied Materials \& Interfaces}, - langid = {english}, - location = {{Washington}}, - number = {3}, - pages = {4034--4042}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {ACS Appl. Mater. Interfaces}, - title = {Machine-{{Learning}}-{{Driven Simulations}} on {{Microstructure}} and {{Thermophysical Properties}} of {{MgCl2}}-{{KCl Eutectic}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {13} -} - -@article{liangMolecularDynamicsSimulations2020, - abstract = {In previous work, molten magnesium chloride has been investigated using first-principles molecular dynamics (FPMD) simulations based on density functional theory (DFT). However, such simulations are computationally intensive and therefore are restricted in terms of simulated size and time. In this work, a machine learning-based deep potential (DP) is trained to accelerate the molecular dynamics simulation of molten magnesium chloride. The trained DP can accurately describe the energies and forces with the prediction errors in energy and force being 1.76 x 10(-3) eV/atom and 4.76 x 10(-2) eV angstrom(-1), respectively. Applying the deep potential molecular dynamics (DPMD) approach, simulations can be performed with more than 1000 atoms, which is infeasible for FPMD simulations. Additionally, the partial radial distribution functions, angle distribution functions, densities, and self-diffusion coefficients predicted by DPMD simulations are also in reasonable agreement with FPMD or experimental results. This work shows that the DP enables higher efficiency and similar accuracy relative to DFT, exhibiting a bright application prospect in modeling molten salt systems.}, - annotation = {WOS:000586457200001}, - author = {Liang, Wenshuo and Lu, Guimin and Yu, Jianguo}, - date = {2020-12}, - doi = {10.1002/adts.202000180}, - journaltitle = {Advanced Theory and Simulations}, - langid = {english}, - location = {{Weinheim}}, - number = {12}, - pages = {2000180}, - publisher = {{Wiley-V C H Verlag Gmbh}}, - shortjournal = {Adv. Theory Simul.}, - title = {Molecular {{Dynamics Simulations}} of {{Molten Magnesium Chloride Using Machine}}-{{Learning}}-{{Based Deep Potential}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {3} -} - -@article{liangTheoreticalPredictionLocal2021a, - abstract = {In this work, the local structure and transport properties of three typical alkali chlorides (LiCl, NaCl, and KCl) were investigated by our newly trained deep potentials (DPs). We extracted datasets from ab initio molecular dynamics (AIMD) calculations and used these to train and validate the DPs. Large-scale and long-time molecular dynamics simulations were performed over a wider range of temperatures than AIMD to confirm the reliability and generality of the DPs. We demonstrated that the generated DPs can serve as a powerful tool for simulating alkali chlorides; the DPs also provide results with accuracy that is comparable to that of AIMD and efficiency that is similar to that of empirical potentials. The partial radial distribution functions and angle distribution functions predicted using the DPs are in close agreement with those derived from AIMD. The estimated densities, self-diffusion coefficients, shear viscosities, and electrical conductivities also matched well with the AIMD and experimental data. This work provides confidence that DPs can be used to explore other systems, including mixtures of chlorides or entirely different salts.}, - author = {Liang, Wenshuo and Lu, Guimin and Yu, Jianguo}, - date = {2021-06-10}, - doi = {10/gmf63v}, - issn = {1005-0302}, - journaltitle = {Journal of Materials Science \& Technology}, - langid = {english}, - pages = {78--85}, - shortjournal = {Journal of Materials Science \& Technology}, - title = {Theoretical Prediction on the Local Structure and Transport Properties of Molten Alkali Chlorides by Deep Potentials}, - url = {https://www.sciencedirect.com/science/article/pii/S1005030220309075}, - urldate = {2021-08-10}, - volume = {75} -} - -@article{liBetterApproximationsHigh2020, - abstract = {Deep neural networks with rectified linear units (ReLU) are getting more and more popular due to their universal representation power and successful applications. Some theoretical progress regarding the approximation power of deep ReLU network for functions in Sobolev space and Korobov space have recently been made by [D. Yarotsky, Neural Network, 94:103-114, 2017] and [H. Montanelli and Q. Du, SIAM J Math. Data Sci., 1:78-92, 2019], etc. In this paper, we show that deep networks with rectified power units (RePU) can give better approximations for smooth functions than deep ReLU networks. Our analysis bases on classical polynomial approximation theory and some efficient algorithms proposed in this paper to convert polynomials into deep RePU networks of optimal size with no approximation error. Comparing to the results on ReLU networks, the sizes of RePU networks required to approximate functions in Sobolev space and Korobov space with an error tolerance epsilon, by our constructive proofs, are in general O(log1/epsilon) times smaller than the sizes of corresponding ReLU networks constructed in most of the existing literature. Comparing to the classical results of Mhaskar [Mhaskar, Adv. Comput. Math. 1:61-80, 1993], our constructions use less number of activation functions and numerically more stable, they can be served as good initials of deep RePU networks and further trained to break the limit of linear approximation theory. The functions represented by RePU networks are smooth functions, so they naturally fit in the places where derivatives are involved in the loss function.}, - annotation = {WOS:000501534800002}, - author = {Li, Bo and Tang, Shanshan and Yu, Haijun}, - date = {2020-02}, - doi = {10.4208/cicp.OA-2019-0168}, - issn = {1815-2406}, - journaltitle = {Communications in Computational Physics}, - langid = {english}, - location = {{Wanchai}}, - number = {2}, - pages = {379--411}, - publisher = {{Global Science Press}}, - shortjournal = {Commun. Comput. Phys.}, - title = {Better {{Approximations}} of {{High Dimensional Smooth Functions}} by {{Deep Neural Networks}} with {{Rectified Power Units}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {27} -} - -@article{liCONFORMATIONGUIDEDMOLECULARREPRESENTA2021, - abstract = {Well-designed molecular representations (fingerprints) are vital to combine medical chemistry and deep learning. Whereas incorporating 3D geometry of molecules (i.e. conformations) in their representations seems beneficial, current 3D algorithms are still in infancy. In this paper, we propose a novel molecular representation algorithm which preserves 3D conformations of molecules with a Molecular Hamiltonian Network (HamNet). In HamNet, implicit positions and momentums of atoms in a molecule interact in the Hamiltonian Engine following the discretized Hamiltonian equations. These implicit coordinations are supervised with real conformations with translation- \& rotation-invariant losses, and further used as inputs to the Fingerprint Generator, a message-passing neural network. Experiments show that the Hamiltonian Engine can well preserve molecular conformations, and that the fingerprints generated by HamNet achieve stateof-the-art performances on MoleculeNet, a standard molecular machine learning benchmark.}, - author = {Li, Ziyao and Yang, Shuwen and Song, Guojie and Cai, Lingsheng}, - date = {2021}, - langid = {english}, - pages = {11}, - title = {{{CONFORMATION}}-{{GUIDED MOLECULAR REPRESENTA}}- {{TION WITH HAMILTONIAN NEURAL NETWORKS}}} -} - -@article{liDevelopmentRobustNeuralnetwork2021a, - abstract = {Molten salts are a promising class of ionic liquids for clean energy applications, such as nuclear and solar energy. However, efficient and accurate evaluation of salt properties from a fundamental, microscopic perspective remains a challenge. Here, we apply artificial neural networks to atomistic modeling of molten NaCl to accurately reproduce the properties from ab initio quantum mechanical calculations based on density functional theory (DFT). The obtained neural network interatomic potential (NNIP) effectively captures the effects of both long-range and short-range interactions, which are crucial for modeling ionic liquids. Extensive validations suggest that the NNIP is capable of predicting the structural, thermophysical, and transport properties of molten NaCl as well as properties of crystalline NaCl, demonstrating near-DFT accuracy and 10(3) x higher efficiency in atomistic simulations. This application of NNIP suggests a paradigm shift from empirical/semiempirical/ab initio approaches to an efficient and accurate machine learning scheme in molten salt modeling.}, - annotation = {WOS:000658764300009}, - author = {Li, Qing-Jie and Kucukbenli, Emine and Lam, Stephen and Khaykovich, Boris and Kaxiras, Efthimios and Li, Ju}, - date = {2021-03-24}, - doi = {10.1016/j.xcrp.2021.100359}, - issn = {2666-3864}, - journaltitle = {Cell Reports Physical Science}, - langid = {english}, - location = {{Amsterdam}}, - number = {3}, - pages = {100359}, - publisher = {{Elsevier}}, - shortjournal = {Cell Rep. Phys. Sci.}, - title = {Development of Robust Neural-Network Interatomic Potential for Molten Salt}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {2} -} - -@article{liEffectLocalStructural2020, - abstract = {Lithium-ion batteries with amorphous silicon (a-Si) anodes exhibit very high theoretical energy capacity, with the lithium kinetic transport having the most crucial effect on the battery performance. In this study, the lithium diffusion pathways in a series of large-scale a-Si models (512 atoms) with various extents of structural order are calculated using the machine learning interatomic potential. Then, the Li diffusion behavior in different atomistic environments is estimated from the transient state theory. The Li diffusion activation energy is observed to be lower (higher) in an ordered (disordered) local environment. The activation energy varies within the range of 1.21-1.46 eV, which agrees well with experimental measurements, 1.38-1.46 eV. Our simulations also show that Li diffusion is enhanced at higher Li concentration, which is consistent with experimental observations. The effects of structural disorder and Li concentration can be explained by the "trap" mechanism. Finally, we show that the sources of Li diffusion traps are dangling bonds and large voids in the a-Si matrix with the help of first-principles calculations. Our work provides insight into the Li diffusion mechanism, which is beneficial for improving the performance of a-Si anodes for lithium-ion batteries. In addition, we demonstrate the significant dependence of the ion transport behavior on the local atomic environment, which will be useful for future theoretical studies of technologically important amorphous materials beyond Si.}, - author = {Li, Wenwen and Ando, Yasunobu}, - date = {2020-04}, - doi = {10.1103/physrevmaterials.4.045602}, - journaltitle = {Physical Review Materials}, - number = {4}, - publisher = {{American Physical Society}}, - title = {Effect of Local Structural Disorder on Lithium Diffusion Behavior in Amorphous Silicon}, - volume = {4} -} - -@article{liHamNetConformationGuidedMolecular2021, - abstract = {Well-designed molecular representations (fingerprints) are vital to combine medical chemistry and deep learning. Whereas incorporating 3D geometry of molecules (i.e. conformations) in their representations seems beneficial, current 3D algorithms are still in infancy. In this paper, we propose a novel molecular representation algorithm which preserves 3D conformations of molecules with a Molecular Hamiltonian Network (HamNet). In HamNet, implicit positions and momentums of atoms in a molecule interact in the Hamiltonian Engine following the discretized Hamiltonian equations. These implicit coordinations are supervised with real conformations with translation- \& rotation-invariant losses, and further used as inputs to the Fingerprint Generator, a message-passing neural network. Experiments show that the Hamiltonian Engine can well preserve molecular conformations, and that the fingerprints generated by HamNet achieve stateof-the-art performances on MoleculeNet, a standard molecular machine learning benchmark.}, - archiveprefix = {arXiv}, - author = {Li, Ziyao and Yang, Shuwen and Song, Guojie and Cai, Lingsheng}, - date = {2021-05-08}, - eprint = {2105.03688}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: in ICLR-2021 (poster)}, - primaryclass = {physics}, - shorttitle = {{{HamNet}}}, - title = {{{HamNet}}: {{Conformation}}-{{Guided Molecular Representation}} with {{Hamiltonian Neural Networks}}}, - url = {http://arxiv.org/abs/2105.03688}, - urldate = {2021-08-11} -} - -@article{liIntroducingBlockDesign2021a, - abstract = {The number of states required for describing a many-body quantum system increases exponentially with the number of particles; thus, it is time- and effort-consuming to exactly calculate molecular properties. Herein, we propose a deep learning algorithm named block-based graph neural network (BGNN) as an approximate solution. The algorithm can be understood as a representation learning process to extract useful interactions between a target atom and its neighboring atomic groups. Compared to other graph model variants, BGNN achieved the smallest mean absolute errors in most tasks on two large molecular datasets, QM9 and Alchemy. Our advanced machine learning method exhibits general applicability and can be readily employed for bioactivity prediction and other tasks relevant to drug discovery and materials design.}, - annotation = {WOS:000641316100001}, - author = {Li, Yuquan and Li, Pengyong and Yang, Xing and Hsieh, Chang-Yu and Zhang, Shengyu and Wang, Xiaorui and Lu, Ruiqiang and Liu, Huanxiang and Yao, Xiaojun}, - date = {2021-06-15}, - doi = {10.1016/j.cej.2021.128817}, - issn = {1385-8947}, - journaltitle = {Chemical Engineering Journal}, - langid = {english}, - location = {{Lausanne}}, - pages = {128817}, - publisher = {{Elsevier Science Sa}}, - shortjournal = {Chem. Eng. J.}, - title = {Introducing Block Design in Graph Neural Networks for Molecular Properties Prediction}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {414} -} - -@article{liMachineLearningComputational2020, - abstract = {The goal of many computational physicists and chemists is the ability to bridge the gap between atomistic length scales of about a few multiples of an Ångström (Å), i. e., 10−10 m, and meso- or macroscopic length scales by virtue of simulations. The same applies to timescales. Machine learning techniques appear to bring this goal into reach. This work applies the recently published on-the-fly machine-learned force field techniques using a variant of the Gaussian approximation potentials combined with Bayesian regression and molecular dynamics as efficiently implemented in the Vienna ab initio simulation package, VASP. The generation of these force fields follows active-learning schemes. We apply these force fields to simple oxides such as MgO and more complex reducible oxides such as iron oxide, examine their generalizability, and further increase complexity by studying water adsorption on these metal oxide surfaces. We successfully examined surface properties of pristine and reconstructed MgO and Fe3O4 surfaces. However, the accurate description of water–oxide interfaces by machine-learned force fields, especially for iron oxides, remains a field offering plenty of research opportunities.}, - author = {Li, Xiaoke and Paier, Wolfgang and Paier, Joachim}, - date = {2020-11-30}, - doi = {10/ghnggc}, - eprint = {33425857}, - eprinttype = {pmid}, - issn = {2296-2646}, - journaltitle = {Frontiers in Chemistry}, - pages = {601029}, - pmcid = {PMC7793815}, - shortjournal = {Front Chem}, - shorttitle = {Machine {{Learning}} in {{Computational Surface Science}} and {{Catalysis}}}, - title = {Machine {{Learning}} in {{Computational Surface Science}} and {{Catalysis}}: {{Case Studies}} on {{Water}} and {{Metal}}–{{Oxide Interfaces}}}, - url = {https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7793815/}, - urldate = {2021-08-11}, - volume = {8} -} - -@article{liMultilevelFineTuningClosing2021a, - abstract = {In scientific machine learning, regression networks have been recently applied to approximate solution maps (e.g., the potential-ground state map of the Schrodinger equation). In this paper, we aim to reduce the generalization error without spending more time on generating training samples. However, to reduce the generalization error, the regression network needs to be fit on a large number of training samples (e.g., a collection of potential-ground state pairs). The training samples can be produced by running numerical solvers, which takes significant time in many applications. In this paper, we aim to reduce the generalization error without spending more time on generating training samples. Inspired by few-shot learning techniques, we develop the multilevel fine-tuning algorithm by introducing levels of training: we first train the regression network on samples generated at the coarsest grid and then successively fine-tune the network on samples generated at finer grids. Within the same amount of time, numerical solvers generate more samples on coarse grids than on fine grids. We demonstrate a significant reduction of generalization error in numerical experiments on challenging problems with oscillations, discontinuities, or rough coefficients. Further analysis can be conducted in the neural tangent kernel regime, and we provide practical estimators to the generalization error. The number of training samples at different levels can be optimized for the smallest estimated generalization error under the constraint of budget for training data. The optimized distribution of budget over levels provides practical guidance with theoretical insight as in the celebrated multilevel Monte Carlo algorithm.}, - annotation = {WOS:000636051200013}, - author = {Li, Zhihan and Fan, Yuwei and Ying, Lexing}, - date = {2021}, - doi = {10.1137/20M1326404}, - issn = {1540-3459}, - journaltitle = {Multiscale Modeling \& Simulation}, - langid = {english}, - location = {{Philadelphia}}, - number = {1}, - pages = {344--373}, - publisher = {{Siam Publications}}, - shortjournal = {Multiscale Model. Simul.}, - shorttitle = {Multilevel {{Fine}}-{{Tuning}}}, - title = {Multilevel {{Fine}}-{{Tuning}}: {{Closing Generalization Gaps}} in {{Approximation}} of {{Solution Maps Under}} a {{Limited Budget}} for {{Training Data}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {19} -} - -@article{liNeuralCanonicalTransformation2020, - abstract = {Canonical transformation plays a fundamental role in simplifying and solving classical Hamiltonian systems. Intriguingly, it has a natural correspondence to normalizing flows with a symplectic constraint. Building on this key insight, we design a neural canonical transformation approach to automatically identify independent slow collective variables in general physical systems and natural datasets. We present an efficient implementation of symplectic neural coordinate transformations and two ways to train the model based either on the Hamiltonian function or phase-space samples. The learned model maps physical variables onto an independent representation where collective modes with different frequencies are separated, which can be useful for various downstream tasks such as compression, prediction, control, and sampling. We demonstrate the ability of this method first by analyzing toy problems and then by applying it to real-world problems, such as identifying and interpolating slow collective modes of the alanine dipeptide molecule and MNIST database images.}, - annotation = {WOS:000529098600001}, - author = {Li, Shuo-Hui and Dong, Chen-Xiao and Zhang, Linfeng and Wang, Lei}, - date = {2020-04-28}, - doi = {10.1103/PhysRevX.10.021020}, - issn = {2160-3308}, - journaltitle = {Physical Review X}, - langid = {english}, - location = {{College Pk}}, - number = {2}, - pages = {021020}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. X}, - title = {Neural {{Canonical Transformation}} with {{Symplectic Flows}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {10} -} - -@article{linNeuralnetworkBasedFramework2021, - abstract = {Interaction potentials are critical to molecular dynamics simulations on fundamental mechanisms at atomic scales. Combination of well-developed single-element empirical potentials via cross interaction (CI) is an important and effective way to develop alloy embedded-atom method (EAM) potentials. In this work, based on neural-network (NN) models, firstly we proposed a framework to construct CI potential functions via utilizing single-element potentials. The framework contained four steps: (1) extracting characteristic points from single-element potential functions, (2) constructing CI functions by cubic spline interpolation, (3) evaluating the accuracy of CI functions by referring to first-principle (FP) data, and (4) searching for reasonable CI functions via NN models. Then with this framework, we developed a Zr-Nb alloy CI potential utilizing the MA-III (pure Zr potential developed by Mendelev and Ackland in 2007) and the Fellinger, Park and Wilkins (FPW) (pure Nb potential developed by FPW in 2010) potentials as single-element parts. The calculated results with this Zr-Nb alloy potential showed that: (1) the newly developed CI potential functions could simultaneously present the potential-function features of Zr and Nb; (2) the normalized energy-volume curves of L1(2) Zr3Nb, B2 ZrNb and L1(2) ZrNb3 calculated by this CI potential reasonably agreed with FP results; (3) the referred MA-III Zr and FPW Nb potentials can satisfactorily reproduce the priority of prismatic slip in Zr and the tension-compression asymmetry of {$<$} 111 \& rang;\{112\} slip in Nb, while other ab initio developed Zr-Nb alloy potentials cannot. Our study indicates that, this NN based framework can take full advantage of single-element potentials, and is very convenient to develop EAM potentials of alloys; moreover, the new-developed Zr-Nb alloy EAM potential can reasonably describe the complicated deformation behaviors in Zr-Nb systems.}, - annotation = {WOS:000596723200001}, - author = {Lin, Bo and Wang, Jincheng and Li, Junjie and Wang, Zhijun}, - date = {2021-02-24}, - doi = {10.1088/1361-648X/abcb69}, - issn = {0953-8984}, - journaltitle = {Journal of Physics-Condensed Matter}, - langid = {english}, - location = {{Bristol}}, - number = {8}, - pages = {084004}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {J. Phys.-Condes. Matter}, - shorttitle = {A Neural-Network Based Framework of Developing Cross Interaction in Alloy Embedded-Atom Method Potentials}, - title = {A Neural-Network Based Framework of Developing Cross Interaction in Alloy Embedded-Atom Method Potentials: Application to {{Zr}}-{{Nb}} Alloy}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {33} -} - -@article{linNumericalMethodsKohnSham2019, - abstract = {Kohn-Sham density functional theory (DFT) is the most widely used electronic structure theory. Despite significant progress in the past few decades, the numerical solution of Kohn-Sham DFT problems remains challenging, especially for large-scale systems. In this paper we review the basics as well as state-of-the-art numerical methods, and focus on the unique numerical challenges of DFT.}, - annotation = {WOS:000491992100004}, - author = {Lin, Lin and Lu, Jianfeng and Ying, Lexing}, - date = {2019}, - doi = {10.1017/S0962492919000047}, - issn = {0962-4929}, - journaltitle = {Acta Numerica}, - langid = {english}, - location = {{New York}}, - pages = {405--539}, - publisher = {{Cambridge Univ Press}}, - shortjournal = {Acta Numer.}, - title = {Numerical Methods for {{Kohn}}-{{Sham}} Density Functional Theory}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {28} -} - -@article{linSearchingConfigurationsUncertainty2021, - abstract = {Neural network (NN) potential energy surfaces (PESs) have been widely used in atomistic simulations with ab initio accuracy. While constructing NN PESs, their training data points are often sampled by molecular dynamics trajectories. This strategy can be however inefficient for reactive systems involving rare events. Here, we develop an uncertainty-driven active learning strategy to automatically and efficiently generate high-dimensional NN-based reactive potentials, taking a gas−surface reaction as an example. The difference between two independent NN models is used as a simple and differentiable uncertainty metric, allowing us to quickly search in the uncertainty space and place new samples at which the PES is less reliable. By interfacing this algorithm with the firstprinciples simulation package, we demonstrate that a globally accurate NN potential of the H2 + Ag(111) system can be constructed with merely ∼150 data points. This PES can be further refined to describe H2 dissociation on Ag(100) by adding ∼130 more configurations on this facet. The entire process is completely automatic and self-terminated once the relative error criterion is fulfilled. Impressively, data points sampled by this uncertainty-driven strategy are substantially fewer than by the traditional trajectory-based sampling. The final NN PES not only converges well the quantum dissociation probability of the molecule but also well-reproduces the phonon properties of the substrate and is capable of describing surface temperature effects. These results show the potential of this active learning approach in developing high-dimensional NN reactive potentials in gas and condensed phases.}, - author = {Lin, Qidong and Zhang, Liang and Zhang, Yaolong and Jiang, Bin}, - date = {2021-05-11}, - doi = {10/gmfw5n}, - issn = {1549-9618, 1549-9626}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - number = {5}, - pages = {2691--2701}, - shortjournal = {J. Chem. Theory Comput.}, - shorttitle = {Searching {{Configurations}} in {{Uncertainty Space}}}, - title = {Searching {{Configurations}} in {{Uncertainty Space}}: {{Active Learning}} of {{High}}-{{Dimensional Neural Network Reactive Potentials}}}, - url = {https://pubs.acs.org/doi/10.1021/acs.jctc.1c00166}, - urldate = {2021-08-11}, - volume = {17} -} - -@article{linUnravellingFastAlkaliIon2021a, - abstract = {Solid-state nuclear magnetic resonance (ssNMR) has received extensive attention in characterizing alkali-ion battery materials because it is highly sensitive for probing the local environment and dynamic information of atoms/ions. However, precise spectral assignment cannot be carried out by conventional DFT for high-rate battery materials at room temperature. Herein, combining DFT calculation of paramagnetic shift and deep potential molecular dynamics (DPMD) simulation to achieve the converged Na+ distribution at hundreds of nanoseconds, we obtain the statistically averaged paramagnetic shift, which is in excellent agreement with ssNMR measurements. Two Na-23 shifts induced by different stacking sequences of transition metal layers are revealed in the fast chemically exchanged NMR spectra of P2-type Na-2/3(Mg1/3Mn2/3)O-2 for the first time. This DPMD simulation auxiliary protocol can be beneficial to a wide range of ssNMR analysis in fast chemically exchanged material systems.}, - annotation = {WOS:000641246200001}, - author = {Lin, Min and Liu, Xiangsi and Xiang, Yuxuan and Wang, Feng and Liu, Yunpei and Fu, Riqiang and Cheng, Jun and Yang, Yong}, - date = {2021-05-25}, - doi = {10.1002/anie.202102740}, - issn = {1433-7851}, - journaltitle = {Angewandte Chemie-International Edition}, - langid = {english}, - location = {{Weinheim}}, - number = {22}, - pages = {12547--12553}, - publisher = {{Wiley-V C H Verlag Gmbh}}, - shortjournal = {Angew. Chem.-Int. Edit.}, - title = {Unravelling the {{Fast Alkali}}-{{Ion Dynamics}} in {{Paramagnetic Battery Materials Combined}} with {{NMR}} and {{Deep}}-{{Potential Molecular Dynamics Simulation}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {60} -} - -@article{liPowerNetEfficientRepresentations2020, - abstract = {Deep neural network with rectified linear units (ReLU) is getting more and more popular recently. However, the derivatives of the function represented by a ReLU network are not continuous, which limit the usage of ReLU network to situations only when smoothness is not required. In this paper, we construct deep neural networks with rectified power units (RePU), which can give better approximations for smooth functions. Optimal algorithms are proposed to explicitly build neural networks with sparsely connected RePUs, which we call PowerNets, to represent polynomials with no approximation error. For general smooth functions, we first project the function to their polynomial approximations, then use the proposed algorithms to construct corresponding PowerNets. Thus, the error of best polynomial approximation provides an upper bound of the best RePU network approximation error. For smooth functions in higher dimensional Sobolev spaces, we use fast spectral transforms for tensor-product grid and sparse grid discretization to get polynomial approximations. Our constructive algorithms show clearly a close connection between spectral methods and deep neural networks: PowerNets with n hidden layers can exactly represent polynomials up to degree s(n), where s is the power of RePUs. The proposed PowerNets have potential applications in the situations where high-accuracy is desired or smoothness is required.}, - annotation = {WOS:000531091900004}, - author = {Li, Bo and Tang, Shanshan and Yu, Haijun}, - date = {2020-06}, - doi = {10.4208/jms.v53n2.20.03}, - issn = {1006-6837}, - journaltitle = {Journal of Mathematical Study}, - langid = {english}, - location = {{Wanchai}}, - number = {2}, - pages = {159--191}, - publisher = {{Global Science Press}}, - shortjournal = {J. Math. Study}, - shorttitle = {{{PowerNet}}}, - title = {{{PowerNet}}: {{Efficient Representations}} of {{Polynomials}} and {{Smooth Functions}} by {{Deep Neural Networks}} with {{Rectified Power Units}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {53} -} - -@article{liTheoreticalStudyNa2021a, - abstract = {Recent discoveries in the antiperovskite-class sodium superionic conductors call for a thorough molecular dynamics (MD) study of sodium ion mobility, but the practical use of MD is often hindered by the accuracy-vs.-efficiency dilemma. Here we applied the recently developed deep potential molecular dynamics (DeePMD) approach to investigate the ion mobility in Na3OBr. With the deep potential model for Na3OBr constructed based on first-principles density-functional theory (DFT) calculations, we directly calculate the Na+ diffusion coefficient at various temperatures, and obtain an activation energy of 0.42-0.43 eV. This in comparison with the 0 K migration barrier (0.41-0.43 eV) suggests that the finite temperature effect is negligible for Na3OBr. The model gives an extrapolated room temperature ionic conductivity of 1 x 10(-4)-2 x 10(-4) mS cm(-1), roughly in the same order of magnitude as the experimental results. We also confirm the proportionality of the diffusion coefficient with respect to the vacancy concentration, and find that the migration barrier is relatively insensitive to the vacancy concentration. This work further demonstrates the promising role of the DeePMD method in the study of the transport properties of solid-state electrolytes.}, - annotation = {WOS:000611857300017}, - author = {Li, Han-Xiao and Zhou, Xu-Yuan and Wang, Yue-Chao and Jiang, Hong}, - date = {2021-01-21}, - doi = {10.1039/d0qi00921k}, - issn = {2052-1553}, - journaltitle = {Inorganic Chemistry Frontiers}, - langid = {english}, - location = {{Cambridge}}, - number = {2}, - pages = {425--432}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Inorg. Chem. Front.}, - title = {Theoretical Study of {{Na}}+ Transport in the Solid-State Electrolyte {{Na3OBr}} Based on Deep Potential Molecular Dynamics}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {8} -} - -@article{liuMachineLearningPhase2021, - abstract = {Derived from phase space expressions of the quantum Liouville theorem, equilibrium continuity dynamics is a category of trajectory-based phase space dynamics methods, which satisfies the two critical fundamental criteria: conservation of the quantum Boltzmann distribution for the thermal equilibrium system and being exact for any thermal correlation functions (even of nonlinear operators) in the classical and harmonic limits. The effective force and effective mass matrix are important elements in the equations of motion of equilibrium continuity dynamics, where only the zeroth term of an exact series expansion of the phase space propagator is involved. We introduce a machine learning approach for fitting these elements in quantum phase space, leading to a much more efficient integration of the equations of motion. Proof-of-concept applications to realistic molecules demonstrate that machine learning phase space dynamics approaches are possible as well as competent in producing reasonably accurate results with a modest computation effort.}, - annotation = {WOS:000649073900013}, - author = {Liu, Xinzijian and Zhang, Linfeng and Liu, Jian}, - date = {2021-05-14}, - doi = {10.1063/5.0046689}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {18}, - pages = {184104}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Machine Learning Phase Space Quantum Dynamics Approaches}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {154} -} - -@article{liUnifiedDeepNeural2020a, - abstract = {Molecular dynamics (MD) simulations have been extensively used to predict thermal properties, but simulating different phases with similar precision using a unified force field is often difficult because of the lack of accurate and transferrable interatomistic potential fields. As a result, this issue has become a major barrier to predicting the phase change of materials and their transport properties with atomistic-level modeling techniques. Recently, machine learning-based algorithms have emerged as promising tools to develop accurate potentials for MD simulations. In this work, we approach the problem of predicting the thermal conductivity of silicon in different phases by performing MD simulations with a deep neural network potential (NNP). This NNP is trained with ab initio data of silicon in the crystalline, liquid, and amorphous phases. The accuracy of our potential is first validated through reproducing the atomistic structures during the phase transition, where other empirical potentials usually fail. The thermal conductivity of different phases is then calculated, showing a good agreement with the experimental results and ab initio calculation results. Our work shows that a unified neural network ebased potential can be a promising tool for studying phase change and thermal transport of materials with high accuracy. (c) 2020 Elsevier Ltd. All rights reserved.}, - annotation = {WOS:000528878800010}, - author = {Li, R. and Lee, E. and Luo, T.}, - date = {2020-03}, - doi = {10.1016/j.mtphys.2020.100181}, - issn = {2542-5293}, - journaltitle = {Materials Today Physics}, - langid = {english}, - location = {{Amsterdam}}, - pages = {100181}, - publisher = {{Elsevier}}, - shortjournal = {Mater. Today Phys.}, - title = {A Unified Deep Neural Network Potential Capable of Predicting Thermal Conductivity of Silicon in Different Phases}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {12} -} - -@article{liuRapidDetectionStrong2020a, - abstract = {Despite its widespread use in chemical discovery, approximate density functional theory (DFT) is poorly suited to many targets, such as those containing open-shell, 3d transition metals that can be expected to have strong multireference (MR) character. For discovery workflows to be predictive, we need automated, low-cost methods that can distinguish the regions of chemical space where DFT should be applied from those where it should not. We curate more than 4800 open-shell transition-metal complexes up to hundreds of atoms in size from prior high-throughput DFT studies and evaluate affordable, finite-temperature DFT fractional occupation number (FON)-based MR diagnostics. We show that intuitive measures of strong correlation (i.e., the HOMO-LUMO gap) are not predictive of MR character as judged by FON-based diagnostics. Analysis of independently trained machine learning (ML) models to predict HOMO-LUMO gaps and FON-based diagnostics reveals differences in the metal and ligand sensitivity of the two quantities. We use our trained ML models to rapidly evaluate MR character over a space of similar to 187000 theoretical complexes, identifying large-scale trends in spin-state-dependent MR character and finding small HOMO-LUMO gap complexes while ensuring low MR character.}, - annotation = {WOS:000577152900021}, - author = {Liu, Fang and Duan, Chenru and Kulik, Heather J.}, - date = {2020-10-01}, - doi = {10.1021/acs.jpclett.0c02288}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {19}, - pages = {8067--8076}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - title = {Rapid {{Detection}} of {{Strong Correlation}} with {{Machine Learning}} for {{Transition}}-{{Metal Complex High}}-{{Throughput Screening}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{liuStructureDynamicsWarm2020, - abstract = {We perform a systematic study on the structure and dynamics of warm dense aluminum (Al) at temperatures ranging from 0.5 to 5.0 eV with molecular dynamics utilizing both density functional theory (DFT) and the deep potential (DP) method. On one hand, unlike the Thomas-Fermi kinetic energy density functional (KEDF), we find that the orbital-free DFT method with the Wang-Teter non-local KEDF yields properties of warm dense Al that agree well with the Kohn-Sham DFT method, enabling accurate orbital-free DFT simulations of warm dense Al at relatively low temperatures. On the other hand, the DP method constructs a deep neural network that has a high accuracy in reproducing short- and long-ranged properties of warm dense Al when compared to the DFT methods. The DP method is orders of magnitudes faster than DFT and is well-suited for simulating large systems and long trajectories to yield accurate properties of warm dense Al. Our results suggest that the combination of DFT methods and the DP model is a powerful tool for accurately and efficiently simulating warm dense matter.}, - annotation = {WOS:000507895000001}, - author = {Liu, Qianrui and Lu, Denghui and Chen, Mohan}, - date = {2020-04-03}, - doi = {10.1088/1361-648X/ab5890}, - issn = {0953-8984}, - journaltitle = {Journal of Physics-Condensed Matter}, - langid = {english}, - location = {{Bristol}}, - number = {14}, - pages = {144002}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {J. Phys.-Condes. Matter}, - shorttitle = {Structure and Dynamics of Warm Dense Aluminum}, - title = {Structure and Dynamics of Warm Dense Aluminum: A Molecular Dynamics Study with Density Functional Theory and Deep Potential}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {32} -} - -@article{liuThermalTransportElectrons2021, - abstract = {We propose an efficient scheme that combines density functional theory (DFT) with deep potentials (DPs), to systematically study convergence issues in the computation of the electronic thermal conductivity of warm dense aluminum (2.7 g/cm3 and temperatures ranging from 0.5 eV to 5.0 eV) with respect to the number of k-points, the number of atoms, the broadening parameter, the exchange-correlation functionals, and the pseudopotentials. Furthermore, we obtain the ionic thermal conductivity using the Green-Kubo method in conjunction with DP molecular dynamics simulations, and we study size effects on the ionic thermal conductivity. This work demonstrates that the proposed method is efficient in evaluating both electronic and ionic thermal conductivities of materials.}, - author = {Liu, Qianrui and Li, Junyi and Chen, Mohan}, - date = {2021-03}, - doi = {10.1063/5.0030123}, - journaltitle = {Matter and Radiation at Extremes}, - number = {2}, - publisher = {{American Institute of Physics Inc.}}, - title = {Thermal Transport by Electrons and Ions in Warm Dense Aluminum: {{A}} Combined Density Functional Theory and Deep Potential Study}, - url = {https://aip.scitation.org/doi/full/10.1063/5.0030123}, - volume = {6} -} - -@article{liuTransferableMultilevelAttention2021, - abstract = {The development of efficient models for predicting specific properties through machine learning is of great importance for the innovation of chemistry and material science. However, predicting global electronic structure properties like Frontier molecular orbital highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels and their HOMO-LUMO gaps from the small-sized molecule data to larger molecules remains a challenge. Here, we develop a multilevel attention neural network, named DeepMoleNet, to enable chemical interpretable insights being fused into multitask learning through (1) weighting contributions from various atoms and (2) taking the atom-centered symmetry functions (ACSFs) as the teacher descriptor. The efficient prediction of 12 properties including dipole moment, HOMO, and Gibbs free energy with in chemical accuracy is achieved by using multiple benchmarks, both at the equilibrium and nonequilibrium geometries, including up to 110,000 records of data in QM9, 400,000 records in MD17, and 280,000 records in ANI-1ccx for random split evaluation. The good transferability for predicting larger molecules outside the training set is demonstrated in both equilibrium QM9 and Alchemy data sets at the density functional theory (DFT) level. Additional tests on nonequilibrium molecular conformations from DFT-based MD17 data set and ANI-1ccx data set with coupled cluster accuracy as well as the public test sets of singlet fission molecules, biomolecules, long oligomers, and protein with up to 140 atoms show reasonable predictions for thermodynamics and electronic structure properties. The proposed multilevel attention neural network is applicable to high-throughput screening of numerous chemical species in both equilibrium and nonequilibrium molecular spaces to accelerate rational designs of drug-like molecules, material candidates, and chemical reactions.}, - annotation = {WOS:000636723700005}, - author = {Liu, Ziteng and Lin, Liqiang and Jia, Qingqing and Cheng, Zheng and Jiang, Yanyan and Guo, Yanwen and Ma, Jing}, - date = {2021-03-22}, - doi = {10.1021/acs.jcim.0c01224}, - issn = {1549-9596}, - journaltitle = {Journal of Chemical Information and Modeling}, - langid = {english}, - location = {{Washington}}, - number = {3}, - pages = {1066--1082}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem Inf. Model.}, - title = {Transferable {{Multilevel Attention Neural Network}} for {{Accurate Prediction}} of {{Quantum Chemistry Properties}} via {{Multitask Learning}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {61} -} - -@article{Loeffler, - abstract = {Artificial Neural Networks (ANNs) for molecular simulations are currently trained by generating large quantities (On the order of 10 4 or greater) of structural data in hopes that the ANN has adequately sampled the energy landscape both near and far-from-equilibrium. This can, however, be a bit prohibitive when it comes to more accurate levels of quantum theory. As such it is desirable to train a model using the absolute minimal data set possible, especially when costs of high-fidelity calculations such as CCSD and QMC are high. Here, we present an Active Learning approach that starts with minimal number of training data points, iteratively samples the energy landscape using nested ensemble Monte Carlo to identify regions of failure and retrains the neural network on-the-fly to improve its performance. We find that this approach is able to train a neural network to reproduce thermodynamic, structure and transport properties of bulk liquid water by sampling less than 300 configurations and their energies. This study reports a new active learning scheme with a promising sampling and training strategy to develop accurate force-fields for molecular simulations using extremely sparse training data sets. The approach is quite generic and can be easily extended to other classes or materials. Abstract Neural Network (NN) based potentials represent flexible alternatives to pre-defined functional forms. Well-trained NN potentials are transferable and provide high level of accuracy on-par with the reference model used for training. Despite their tremendous potentials and interests, there are at least two challenges that need to be addressed-(1) NN models are interpolative and hence trained by generating large quantities (∼ 10 4 or greater) of structural data in hopes that the model has adequately sampled the energy landscape both near and far-from-equilibrium. It is desirable to minimize the number of training data, especially if the underlying reference model is expensive. (2) NN atomistic potentials (like any other classical atomistic model) are limited in the time scales they can access. Coarse-grained NN potentials have emerged as a viable alternative. Here, we address these challenges by introducing an active learning scheme that trains a CG model with minimal amount of training data. Our active learning workflow starts with a sparse training dataset (∼1 to 5 data points) which is continually updated via a Nested Ensemble Monte Carlo scheme that iteratively queries the energy landscape in regions of failure and improves the network performance. We demonstrate that with ∼300 reference data, our AL-NN is able to accurately predict both the energies and the molecular forces of water, within 2 meV/molecule and 40 meV/Å of the reference (coarse-grained bond-order potential) model. The AL-NN water model provides good prediction of several structural, thermodynamic, and temperature dependent properties of liquid water, with values close to those obtained from the reference model. AL-NN also captures the well-known density anomaly of liquid water observed in experiments. While the AL procedure has been demonstrated for training CG models with sparse}, - author = {Loeffler, TD and Patra, TK and H, Chan}, - journaltitle = {pubs.rsc.org}, - title = {Active Learning a Coarse-Grained Neural Network Model for Bulk Water from Sparse Training Data}, - url = {https://pubs.rsc.org/en/content/articlehtml/2020/me/c9me00184k} -} - -@article{Loeffler2020, - abstract = {Small metal clusters are of fundamental scientific interest and of tremendous significance in catalysis. These nanoscale clusters display diverse geometries and structural motifs depending on the cluster size; a knowledge of this size-dependent structural motifs and their dynamical evolution has been of longstanding interest. Given the high computational cost of first-principles calculations, molecular modeling and atomistic simulations such as molecular dynamics (MD) has proven to be an important complementary tool to aid this understanding. Classical MD typically employ predefined functional forms which limits their ability to capture such complex size-dependent structural and dynamical transformation. Neural Network (NN) based potentials represent flexible alternatives and in principle, well-trained NN potentials can provide high level of flexibility, transferability and accuracy on-par with the reference model used for training. A major challenge, however, is that NN models are interpolative and requires large quantities (∼ 10 4 or greater) of training data to ensure that the model adequately samples the energy landscape both near and far-from-equilibrium. A highly desirable goal is minimize the number of training data, especially if the underlying reference model is first-principles based and hence expensive. Here, we introduce an active learning (AL) scheme that trains a NN model on-the-fly with minimal amount of first-principles based training data. Our AL workflow is initiated with a sparse training dataset (∼ 1 to 5 data points) and is updated on-the-fly via a Nested Ensemble Monte Carlo scheme that iteratively queries the energy landscape in regions of failure and updates the training pool to improve the network performance. Using a representative system of gold clusters, we demonstrate that our AL workflow can train a NN with ∼ 500 total reference calculations. Using an extensive DFT test set of ∼ 1100 configurations, we show that our AL-NN is able to accurately predict both the DFT energies and the forces for clusters of a myriad of different sizes. Our NN predictions are within 30 meV/atom and 40 meV/ ˚ A of the reference DFT calculations. Moreover, our AL-NN model also adequately captures the various size-dependent structural and dynamical properties of gold clusters in excellent agreement with DFT calculations and available experiments. We finally show that our AL-NN model also captures bulk properties reasonably well, even though they were not included in the training data.}, - archiveprefix = {arXiv}, - arxivid = {2006.03674v1}, - author = {Loeffler, TD and Manna, S and Patra, TK and H, Chan}, - date = {2020}, - eprint = {2006.03674v1}, - eprinttype = {arxiv}, - journaltitle = {arxiv.org}, - title = {Active Learning a Neural Network Model for Gold Clusters\& Bulk from Sparse First Principles Training Data}, - url = {https://arxiv.org/abs/2006.03674} -} - -@article{Loeffler2020a, - abstract = {Molecular dynamics with predefined functional forms is a popular technique for understanding dynamical evolution of systems. The predefined functional forms impose limits on the physics that can be captured. Artificial neural network (ANN) models have emerged as an attractive flexible alternative to the expensive quantum calculations (e.g., density functional theory) in the area of molecular force-fields. Ideally, if one is able to train a ANN to accurately predict the correct DFT energy and forces for any given structure, they gain the ability to perform molecular dynamics with high accuracy while simultaneously reducing the computation cost in a dramatic fashion. While this goal is very lucrative, neural networks are interpolative and therefore, it is not always clear how one should go about training a neural network to exhaustively fit the entire phase space of a given system. Currently, ANNs are trained by generating large quantities (on the order of 104 or greater) of training data in hopes that the ANN has adequately sampled the energy landscape both near and far-from-equilibrium. This can, however, be a bit prohibitive when it comes to more accurate levels of quantum theory. As such, it is desirable to train a model using the absolute minimal data set possible, especially when costs of high-fidelity calculations such as CCSD and QMC are high. Here, we present an active learning approach that iteratively trains an ANN model to faithfully replicate the coarse-grained energy surface of water clusters using only 426 total structures in its training data. Our active learning workflow starts with a sparse training data set which is continually updated via a Nested Ensemble Monte Carlo scheme that sparsely queries the energy landscape and tests the network performance. Next, the network is retrained with an updated training set that includes failed configurations/energies from previous iteration until convergence is attained. Once trained, we generate an extensive test set of 100 »000 configurations sampled across clusters ranging from 1 to 200 molecules and demonstrate that the trained network adequately reproduces the energies (within mean absolute error (MAE) of 2 meV/molecule) and forces (MAE 40 meV/Å) compared to the reference model. More importantly, the trained ANN model also accurately captures both the structure as well as the free energy as a function of the various cluster sizes. Overall, this study reports a new active learning scheme with promising strategy to develop accurate force-fields for molecular simulations using extremely sparse training data sets.}, - author = {Loeffler, Troy D. and Patra, Tarak K. and Chan, Henry and Cherukara, Mathew and Sankaranarayanan, Subramanian K.R.S.}, - date = {2020-02}, - doi = {10.1021/acs.jpcc.0c00047}, - journaltitle = {Journal of Physical Chemistry C}, - number = {8}, - pages = {4907--4916}, - publisher = {{American Chemical Society}}, - title = {Active Learning the Potential Energy Landscape for Water Clusters from Sparse Training Data}, - volume = {124} -} - -@article{longPDENetLearningPDEs2019a, - abstract = {Partial differential equations (PDEs) are commonly derived based on empirical observations. However, recent advances of technology enable us to collect and store massive amount of data, which offers new opportunities for data-driven discovery of PDEs. In this paper, we propose a new deep neural network, called PDE-Net 2.0, to discover (time-dependent) PDEs from observed dynamic data with minor prior knowledge on the underlying mechanism that drives the dynamics. The design of PDE-Net 2.0 is based on our earlier work [1] where the original version of PDE-Net was proposed. PDE-Net 2.0 is a combination of numerical approximation of differential operators by convolutions and a symbolic multi-layer neural network for model recovery. Comparing with existing approaches, PDE-Net 2.0 has the most flexibility and expressive power by learning both differential operators and the nonlinear response function of the underlying PDE model. Numerical experiments show that the PDE-Net 2.0 has the potential to uncover the hidden PDE of the observed dynamics, and predict the dynamical behavior for a relatively long time, even in a noisy environment. (C) 2019 Elsevier Inc. All rights reserved.}, - annotation = {WOS:000490766100033}, - author = {Long, Zichao and Lu, Yiping and Dong, Bin}, - date = {2019-12-15}, - doi = {10.1016/j.jcp.2019.108925}, - issn = {0021-9991}, - journaltitle = {Journal of Computational Physics}, - langid = {english}, - location = {{San Diego}}, - pages = {108925}, - publisher = {{Academic Press Inc Elsevier Science}}, - shortjournal = {J. Comput. Phys.}, - shorttitle = {{{PDE}}-{{Net}} 2.0}, - title = {{{PDE}}-{{Net}} 2.0: {{Learning PDEs}} from Data with a Numeric-Symbolic Hybrid Deep Network}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {399} -} - -@article{lotPANNAPropertiesArtificial2020, - abstract = {Prediction of material properties from first principles is often a computationally expensive task. Recently, artificial neural networks and other machine learning approaches have been successfully employed to obtain accurate models at a low computational cost by leveraging existing example data. Here, we present a software package "Properties from Artificial Neural Network Architectures'' (PANNA) that provides a comprehensive toolkit for creating neural network models for atomistic systems following the Behler-Parrinello topology. Besides the core routines for neural network training, it includes data parser, descriptor builder for Behler-Parrinello class of symmetry functions and forcefield generator suitable for integration within molecular dynamics packages. PANNA offers a variety of activation and cost functions, regularization methods, as well as the possibility of using fullyconnected networks with custom size for each atomic species. PANNA benefits from the optimization and hardware-flexibility of the underlying TensorFlow engine which allows it to be used on multiple CPU/GPU/TPU systems, making it possible to develop and optimize neural network models based on large datasets. Program summary Program title: PANNA-Properties from Artificial Neural Network Architectures CPC Library link to program files: http://dx.doi.org/10.17632/mcryj6cnnh.1 Licensing provisions: MIT Programming language: Python, C++ Nature of problem: A workflow for machine learning atomistic properties and interatomic potentials using neural networks. Solution method: This package first transforms the user supplied data into pairs of precomputed input (Behler-Parrinello [1] class of symmetry functions) and target output (energy and forces) for the neural network model. The data are then packed to enable efficient reading. A user-friendly interface to TensorFlow [2] is provided to instantiate and train neural network models with varying architectures within Behler-Parrinello topology and with varying training schedules. The training can be monitored and validated with the provided tools. The derivative of the target output with respect to the input can also be used jointly in training, e.g. in the case of energy and force training. The interface with molecular dynamics codes such as LAMMPS [3] allows the neural network model to be used as an interatomic potential. Additional comments including restrictions and unusual features: The underlying neural network training engine, TensorFlow, is a prerequisite of PANNA. While there is a special LAMMPS integration performed via a patch distributed within PANNA, the network potentials can be deposited into OpenKIM [4] database and can be used with a wide range of molecular dynamics codes. The package allows different network architectures to be used for each atomic species, with different trainability setting for each network layer. It provides tools of exchanging weights between atomic species, and provides the option of building a Radial Basis Function network. The software is parallelized to take advantage of hardware architectures with multiple CPU/GPU/TPUs. (C) 2020 Elsevier B.V. All rights reserved.}, - annotation = {WOS:000590251400006}, - author = {Lot, Ruggero and Pellegrini, Franco and Shaidu, Yusuf and Kucukbenli, Emine}, - date = {2020-11}, - doi = {10.1016/j.cpc.2020.107402}, - issn = {0010-4655}, - journaltitle = {Computer Physics Communications}, - langid = {english}, - location = {{Amsterdam}}, - pages = {107402}, - publisher = {{Elsevier}}, - shortjournal = {Comput. Phys. Commun.}, - shorttitle = {{{PANNA}}}, - title = {{{PANNA}}: {{Properties}} from {{Artificial Neural Network Architectures}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {256} -} - -@article{Lu2019, - abstract = {The microstructures of materials determine their macroscopic properties. The traditional bottom up multi-scale approach provides a general strategy for studying the relationship between microstructures and physical properties. However, there are still many difficulties in microscopic, mesoscopic and macroscopic modeling of materials, and the bridging of different scale models is extremely challenging. With the advancement of computing power in Moore's Law, and the explosive development of artificial intelligence, especially deep learning, data-driven methods are commonly used, and in crystal structure prediction, stability analysis, equation of states, optical properties, chemical synthesis, etc. have achieved good application results. The fast calculation speed and reliable prediction capabilities of deep learning can greatly improve the efficiency of material simulation. Its wide applicability provides new research ideas for some traditional problems in material microstructures and multi-scale simulations. It is expected to promote the study of material microstructures and physical properties, and to provide new research directions for modelling macroscopic physical properties based on micro-mesoscopic mechanism and prediction of material properties to meet engineering application requirements. This review article will briefly introduce the basic principles of deep learning and main types of commonly used neural networks, outline the main methods of material microstructures and multi-scale modeling, and then introduce the recent progress of deep learning method in the study of material microstruc-ture and physical properties, and review the developments and prospects of deep learning method in the field of multi-scale simulation of materials. Keywords Deep Learning, Material Microstructures and Materials Properties, Multiscale Simulation 深度学习：材料微结构与物性研究中的新动力 卢 果，段素青}, - author = {Lu, G and Duan, S}, - date = {2019}, - doi = {10.12677/mp.2019.96026}, - journaltitle = {Modern Physics 现代物理}, - number = {6}, - pages = {263--276}, - title = {Deep Learning: {{New}} Engine for the Study of Material Microstructures and Physical Properties}, - url = {https://pdf.hanspub.org/MP20190600000_96988416.pdf}, - volume = {2019} -} - -@article{luDatasetConstructionExplore2021, - abstract = {A dataset is the basis of deep learning model development, and the success of deep learning models heavily relies on the quality and size of the dataset. In this work, we present a new data preparation protocol and build a large fragment-based dataset Frag20, which consists of optimized 3D geometries and calculated molecular properties from Merck molecular force field (MMFF) and DFT at the B3LYP/6-31G* level of theory for more than half a million molecules composed of H, B, C, O, N, F, P, S, Cl, and Br with no larger than 20 heavy atoms. Based on the new dataset, we develop robust molecular energy prediction models using a simplified PhysNet architecture for both DFT-optimized and MMFF-optimized geometries, which achieve better than or close to chemical accuracy (1 kcal/mol) on multiple test sets, including CSD20 and Plati20 based on experimental crystal structures.}, - annotation = {WOS:000636723700007}, - author = {Lu, Jianing and Xia, Song and Lu, Jieyu and Zhang, Yingkai}, - date = {2021-03-22}, - doi = {10.1021/acs.jcim.1c00007}, - issn = {1549-9596}, - journaltitle = {Journal of Chemical Information and Modeling}, - langid = {english}, - location = {{Washington}}, - number = {3}, - pages = {1095--1104}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem Inf. Model.}, - title = {Dataset {{Construction}} to {{Explore Chemical Space}} with {{3D Geometry}} and {{Deep Learning}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {61} -} - -@article{luDeepPotentialMolecular2021a, - abstract = {We present the GPU version of DeePMD-kit, which, upon training a deep neural network model using ab initio data, can drive extremely large-scale molecular dynamics (MD) simulation with ab initio accuracy. Our tests show that for a water system of 12, 582, 912 atoms, the GPU version can be 7 times faster than the CPU version under the same power consumption. The code can scale up to the entire Summit supercomputer. For a copper system of 113, 246, 208 atoms, the code can perform one nanosecond MD simulation per day, reaching a peak performance of 86 PFLOPS (43\% of the peak). Such unprecedented ability to perform MD simulation with ab initio accuracy opens up the possibility of studying many important issues in materials and molecules, such as heterogeneous catalysis, electrochemical cells, irradiation damage, crack propagation, and biochemical reactions. (C) 2020 Elsevier B.V. All rights reserved.}, - annotation = {WOS:000599925900017}, - author = {Lu, Denghui and Wang, Han and Chen, Mohan and Lin, Lin and Car, Roberto and E, Weinan and Jia, Weile and Zhang, Linfeng}, - date = {2021-02}, - doi = {10.1016/j.cpc.2020.107624}, - issn = {0010-4655}, - journaltitle = {Computer Physics Communications}, - langid = {english}, - location = {{Amsterdam}}, - pages = {107624}, - publisher = {{Elsevier}}, - shortjournal = {Comput. Phys. Commun.}, - title = {Deep {{Potential Molecular Dynamics}} Simulation of 100 Million Atoms with Ab Initio Accuracy}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {259} -} - -@article{luDeepPotentialMolecular2021a, - abstract = {We present the GPU version of DeePMD-kit, which, upon training a deep neural network model using ab initio data, can drive extremely large-scale molecular dynamics (MD) simulation with ab initio accuracy. Our tests show that for a water system of 12, 582, 912 atoms, the GPU version can be 7 times faster than the CPU version under the same power consumption. The code can scale up to the entire Summit supercomputer. For a copper system of 113, 246, 208 atoms, the code can perform one nanosecond MD simulation per day, reaching a peak performance of 86 PFLOPS (43\% of the peak). Such unprecedented ability to perform MD simulation with ab initio accuracy opens up the possibility of studying many important issues in materials and molecules, such as heterogeneous catalysis, electrochemical cells, irradiation damage, crack propagation, and biochemical reactions. (C) 2020 Elsevier B.V. All rights reserved.}, - annotation = {WOS:000599925900017}, - author = {Lu, Denghui and Wang, Han and Chen, Mohan and Lin, Lin and Car, Roberto and E, Weinan and Jia, Weile and Zhang, Linfeng}, - date = {2021-02}, - doi = {10.1016/j.cpc.2020.107624}, - issn = {0010-4655}, - journaltitle = {Computer Physics Communications}, - langid = {english}, - location = {{Amsterdam}}, - pages = {107624}, - publisher = {{Elsevier}}, - shortjournal = {Comput. Phys. Commun.}, - title = {Deep {{Potential Molecular Dynamics}} Simulation of 100 Million Atoms with Ab Initio Accuracy}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {259} -} - -@article{luDPTrainThen2021, - abstract = {Machine learning based interatomic potential energy surface (PES) models are revolutionizing the field of molecular modeling. However, although much faster than electronic structure schemes, these models suffer from a lower efficiency as compared to typical empirical force fields due to more sophisticated computations involved. Herein, we report a model compression scheme for boosting the performance of the Deep Potential (DP) model, a deep learning based PES model. This scheme, we call DP Compress, is an efficient post-processing step after the training of DP models (DP Train). DP Compress combines several DP-specific compression techniques, which typically speed up DP-based molecular dynamics simulations by an order of magnitude faster, and consume an order of magnitude less memory. We demonstrate that DP Compress is sufficiently accurate by testing a variety of physical properties of Cu, H2O, and Al-Cu-Mg systems. DP Compress applies to both CPU and GPU machines and is publicly available at https://github.com/deepmodeling/deepmd-kit.}, - archiveprefix = {arXiv}, - arxivid = {2107.02103v1}, - author = {Lu, D and Jiang, W and Chen, Y and Zhang, L and Jia, W and Wang, H}, - date = {2021}, - eprint = {2107.02103v1}, - eprinttype = {arxiv}, - isbn = {2107.02103v1}, - journaltitle = {arxiv.org}, - title = {{{DP}} Train, Then {{DP}} Compress: {{Model}} Compression in Deep Potential Molecular Dynamics}, - url = {https://arxiv.org/abs/2107.02103} -} - -@article{lunghiUnifiedPictureCovalent2019, - abstract = {Computational studies of chemical processes taking place over extended size and time scales are inaccessible by electronic structure theories and can be tackled only by atomistic models such as force fields. These have evolved over the years to describe the most diverse systems. However, as we improve the performance of a force field for a particular physical/chemical situation, we are also moving away from a unified description. Here, we demonstrate that a unified picture of the covalent bond is achievable within the framework of machine learning-based force fields. Ridge regression, together with a representation of the atomic environment in terms of bispectrum components, can be used to map a general potential energy surface for molecular systems at chemical accuracy. This protocol sets the ground for the generation of an accurate and universal class of potentials for both organic and organometallic compounds with no specific assumptions on the chemistry involved.}, - annotation = {WOS:000470125000103}, - author = {Lunghi, Alessandro and Sanvito, Stefano}, - date = {2019-05}, - doi = {10.1126/sciadv.aaw2210}, - issn = {2375-2548}, - journaltitle = {Science Advances}, - langid = {english}, - location = {{Washington}}, - number = {5}, - pages = {eaaw2210}, - publisher = {{Amer Assoc Advancement Science}}, - shortjournal = {Sci. Adv.}, - shorttitle = {A Unified Picture of the Covalent Bond within Quantum-Accurate Force Fields}, - title = {A Unified Picture of the Covalent Bond within Quantum-Accurate Force Fields: {{From}} Organic Molecules to Metallic Complexes' Reactivity}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {5} -} - -@article{luoAnomalousBehaviorViscosity2021, - abstract = {Silicate melts have served as transport agents in the chemical and thermal evolution of Earth. Molecular dynamics simulations based on a deep neural network potential trained by ab initio data show that the viscosity of MgSiO3 melt decreases with increasing pressure at low pressures (up to ∼6 GPa) before it starts to increase with further compression. The melt electrical conductivity also behaves anomalously; first increasing and then decreasing with pressure. The melt accumulation implied by the viscosity turnover at ∼23 GPa along mantle liquidus offers an explanation for the low-velocity zone at the 660-km discontinuity. The increase in electrical conductivity up to ∼50 GPa may contribute to the steep rise of Earth's electrical conductivity profiles derived from magnetotelluric observations. Our results also suggest that small fraction of melts could give rise to detectable bulk conductivity in deeper parts of the mantle.}, - annotation = {\_eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2021GL093573}, - author = {Luo, Haiyang and Karki, Bijaya B. and Ghosh, Dipta B. and Bao, Huiming}, - date = {2021-06-16}, - doi = {10/gkrt5v}, - issn = {1944-8007}, - journaltitle = {Geophysical Research Letters}, - langid = {english}, - number = {13}, - pages = {e2021GL093573}, - title = {Anomalous {{Behavior}} of {{Viscosity}} and {{Electrical Conductivity}} of {{MgSiO3 Melt}} at {{Mantle Conditions}}}, - url = {https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2021GL093573}, - urldate = {2021-08-11}, - volume = {48} -} - -@article{luoDeepNeuralNetwork2021, - abstract = {Diffusional isotope fractionation has been widely used to explain lithium (Li) isotope variations in minerals and rocks. Isotopic mass dependence of Li diffusion can be empirically expressed as D7LiD6Li=67β, where D is the diffusivity of a Li isotope. The knowledge about temperature and compositional dependence of the β factor which is essential for understanding diffusion profiles and mechanisms remains unclear. Based on the potential energy and interatomic forces generated by deep neural networks trained with ab initio data, we performed deep potential molecular dynamics (DPMD) simulations of several Li pseudo-isotopes (with mass = 2, 7, 21, 42 g/mol) in albite, hydrous albite, and model basalt melts to evaluate the β factor. Our calculated diffusivities for 7Li in albite and model basalt melts at 1800 K compare well with experimental results. We found that β in albite melt decreases from 0.267±0.006 at 4000 K to 0.225±0.004 at 1800 K. The presence of water appears to slightly weaken the temperature dependence of β, with β decreasing from 0.250±0.012 to 0.228±0.031 in hydrous albite melt. The calculated β in model basalt melt takes much smaller values, decreasing from 0.215±0.006 at 4000 K to 0.132±0.015 at 1800 K. Our prediction of β in albite and hydrous albite melts is in good agreement with experimental data. More importantly, our results suggest that Li isotope diffusion in silicate melts is strongly dependent on melt composition. The temperature and compositional effects on β can be qualitatively explained in terms of ionic porosity and the coupled relationship between Li diffusion and the mobility of the silicate melt network. Two types of diffusion experiments are suggested to test our predicted temperature and compositional dependence of β. This study shows that DPMD is a promising tool to simulate the diffusion of elements and isotopes in silicate melts.}, - author = {Luo, Haiyang and Karki, Bijaya B. and Ghosh, Dipta B. and Bao, Huiming}, - date = {2021-06-15}, - doi = {10/gmf625}, - issn = {0016-7037}, - journaltitle = {Geochimica et Cosmochimica Acta}, - langid = {english}, - pages = {38--50}, - shortjournal = {Geochimica et Cosmochimica Acta}, - title = {Deep Neural Network Potentials for Diffusional Lithium Isotope Fractionation in Silicate Melts}, - url = {https://www.sciencedirect.com/science/article/pii/S0016703721002052}, - urldate = {2021-08-10}, - volume = {303} -} - -@article{luPredictingMolecularEnergy2019a, - abstract = {The use of neural networks to predict molecular properties calculated from high level quantum mechanical calculations has made significant advances in recent years, but most models need input geometries from DFT optimizations which limit their applicability in practice. In this work, we explored how machine learning can be used to predict molecular atomization energies and conformation stability using optimized geometries from Merck Molecular Force Field (MMFF). On the basis of the recently introduced deep tensor neural network (DTNN) approach, we first improved its training efficiency and performed an extensive search of its hyperparameters, and developed a DTNN\_7ib model which has a test accuracy of 0.34 kcal/mol mean absolute error (MAE) on QM9 data set. Then using atomic vector representations in the DTNN\_7ib model, we employed transfer learning (TL) strategy to train readout layers on the QM9(M) data set, in which QM properties are the same as in QM9 [calculated at the B3LYP/6-31G(2df,p) level] while molecular geometries are corresponding local minima optimized with MMFF94 force field. The developed TL\_QM9(M) model can achieve an MAE of 0.79 kcal/mol using MMFF optimized geometries. Furthermore, we demonstrated that the same transfer learning strategy with the same atomic vector representation can be used to develop a machine learning model that can achieve an MAE of 0.51 kcal/mol in molecular energy prediction using MMFF geometries for an eMol9\_C-M conformation data set, which consists of 9959 molecules and 88 234 conformations with energies calculated at the B3LYP/6-31G* level. Our results indicate that DFT-level accuracy of molecular energy prediction can be achieved using force-field optimized geometries and atomic vector representations learned from deep tensor neural network, and integrated molecular modeling and machine learning would be a promising approach to develop more powerful computational tools for molecular conformation analysis.}, - annotation = {WOS:000475409000019}, - author = {Lu, Jianing and Wang, Cheng and Zhang, Yingkai}, - date = {2019-07}, - doi = {10.1021/acs.jctc.9b00001}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {7}, - pages = {4113--4121}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {Predicting {{Molecular Energy Using Force}}-{{Field Optimized Geometries}} and {{Atomic Vector Representations Learned}} from an {{Improved Deep Tensor Neural Network}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {15} -} - -@article{Lye2019, - abstract = {Many large scale problems in computational fluid dynamics such as uncertainty quantification, Bayesian inversion, data assimilation and PDE constrained optimization are considered very challenging computationally as they require a large number of expensive (forward) numerical solutions of the corresponding PDEs. We propose a machine learning algorithm, based on deep artificial neu-ral networks, that predicts the underlying input parameters to observable map from a few training samples (computed realizations of this map). By a judicious combination of theoretical arguments and empirical observations, we find suitable network architectures and training hyperparameters that result in robust and efficient neural network approximations of the parameters to observable map. Numerical experiments are presented to demonstrate low prediction errors for the trained network networks, even when the network has been trained with a few samples, at a computational cost which is several orders of magnitude lower than the underlying PDE solver. Moreover, we combine the proposed deep learning algorithm with Monte Carlo (MC) and Quasi-Monte Carlo (QMC) methods to efficiently compute uncertainty propagation for nonlinear PDEs. Under the assumption that the underlying neural networks generalize well, we prove that the deep learning MC and QMC algorithms are guaranteed to be faster than the baseline (quasi-) Monte Carlo methods. Numerical experiments demonstrating one to two orders of magnitude speed up over baseline QMC and MC algorithms, for the intricate problem of computing probability distributions of the observable, are also presented.}, - archiveprefix = {arXiv}, - arxivid = {1903.03040v2}, - author = {Lye, KO and Mishra, S and Physics, D Ray - Journal of Computational and 2020, undefined}, - date = {2019}, - eprint = {1903.03040v2}, - eprinttype = {arxiv}, - journaltitle = {Elsevier}, - title = {Deep Learning Observables in Computational Fluid Dynamics}, - url = {https://www.sciencedirect.com/science/article/pii/S0021999120301133} -} - -@article{mailoaFastNeuralNetwork2019a, - abstract = {Neural network force field (NNFF) is a method for performing regression on atomic structure-force relationships, bypassing the expensive quantum mechanics calculations that prevent the execution of long ab initio quality molecular dynamics (MD) simulations. However, most NNFF methods for complex multi-element atomic systems indirectly predict atomic force vectors by exploiting just atomic structure rotation-invariant features and network-feature spatial derivatives, which are computationally expensive. Here, we show a staggered NNFF architecture that exploits both rotation-invariant and -covariant features to directly predict atomic force vectors without using spatial derivatives, and we demonstrate 2.2x NNFF-MD acceleration over a state-of-the-art C++ engine using a Python engine. This fast architecture enables us to develop NNFF for complex ternary- and quaternary-element extended systems composed of long polymer chains, amorphous oxide and surface chemical reactions. The rotation-invariant-covariant architecture described here can also directly predict complex covariant vector outputs from local environments, in other domains beyond computational material science. Neural network force fields promise to bypass the computationally expensive quantum mechanical calculations typically required to investigate complex materials, such as lithium-ion batteries. Mailoa et al. accelerate these approaches with an architecture that exploits both rotation-invariant and -covariant features separately.}, - annotation = {WOS:000571255900005}, - author = {Mailoa, Jonathan P. and Kornbluth, Mordechai and Batzner, Simon and Samsonidze, Georgy and Lam, Stephen T. and Vandermause, Jonathan and Ablitt, Chris and Molinari, Nicola and Kozinsky, Boris}, - date = {2019-10}, - doi = {10.1038/s42256-019-0098-0}, - journaltitle = {Nature Machine Intelligence}, - langid = {english}, - location = {{London}}, - number = {10}, - pages = {471--479}, - publisher = {{Springernature}}, - shortjournal = {Nat. Mach. Intell.}, - title = {A Fast Neural Network Approach for Direct Covariant Forces Prediction in Complex Multi-Element Extended Systems}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {1} -} - -@article{maityEvaluationExperimentalAlkali2019, - abstract = {We applied the domain based local pair natural orbital coupled cluster approach with single, double, and perturbative triple excitations, DLPNO-CCSD(T), to rationalize more than 130 experimental bond dissociation enthalpies collected in the work of Rodgers and Armentrout [Chem. Rev. 116, 5642-5687 (2016)] and involving alkali metal cations and versatile neutral organic and inorganic ligands ranging from common solvents to amino acids. In general, a remarkable agreement has been obtained between predicted and experimental alkali metal ion-ligand noncovalent bond strengths, highlighting a high degree of reliability of data assembled by Rodgers and Armentrout. In the case of some inconsistent experimental data given for some species, we pointed to a number for which best agreement with DLPNO-CCSD(T) calculations has been achieved. In addition, we refined a couple of Delta H-0 for which DLPNO-CCSD(T) values turned out to be significantly different from their experimental counterparts. We suggest an application of the DLPNO-CCSD(T) to derive the reference values to train/validate force field and neural network methods to be further applied in molecular dynamic simulations to unravel the mechanisms in biological systems and alkali metal ion batteries.}, - annotation = {WOS:000474214600014}, - author = {Maity, Bholanath and Minenkov, Yury and Cavallo, Luigi}, - date = {2019-07-07}, - doi = {10.1063/1.5099580}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {1}, - pages = {014301}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Evaluation of Experimental Alkali Metal Ion-Ligand Noncovalent Bond Strengths with {{DLPNO}}-{{CCSD}}({{T}}) Method}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {151} -} - -@article{mangoldTransferabilityNeuralNetwork2020, - abstract = {Germanium manganese compounds exhibit a variety of stable and metastable phases with different stoichiometry. These materials entail interesting electronic, magnetic and thermal properties both in their bulk form and as heterostructures. Here we develop and validate a transferable machine learning potential, based on the high-dimensional neural network formalism, to enable the study of Mn\$\_x\$Ge\$\_y\$ materials over a wide range of compositions. We show that a neural network potential fitted on a minimal training set reproduces successfully the structural and vibrational properties and the thermal conductivity of systems with different local chemical environments, and it can be used to predict phononic effects in nanoscale heterostructures.}, - archiveprefix = {arXiv}, - author = {Mangold, Claudia and Chen, Shunda and Barbalinardo, Giuseppe and Behler, Joerg and Pochet, Pascal and Termentzidis, Konstantinos and Han, Yang and Chaput, Laurent and Lacroix, David and Donadio, Davide}, - date = {2020-06-28}, - doi = {10/gg7jww}, - eprint = {2005.09591}, - eprinttype = {arxiv}, - issn = {0021-8979, 1089-7550}, - journaltitle = {Journal of Applied Physics}, - langid = {english}, - note = {Comment: 29 pages, 9 figures}, - number = {24}, - pages = {244901}, - shortjournal = {Journal of Applied Physics}, - shorttitle = {Transferability of Neural Network Potentials for Varying Stoichiometry}, - title = {Transferability of Neural Network Potentials for Varying Stoichiometry: Phonons and Thermal Conductivity of {{Mn}}\$\_x\${{Ge}}\$\_y\$ Compounds}, - url = {http://arxiv.org/abs/2005.09591}, - urldate = {2021-08-11}, - volume = {127} -} - -@article{marchandMachineLearningMetallurgy2020, - abstract = {High-strength metal alloys achieve their performance via careful control of precipitates and solutes. The nucleation, growth, and kinetics of precipitation, and the resulting mechanical properties, are inherently atomic scale phenomena, particularly during early-stage nucleation and growth. Atomistic modeling using interatomic potentials is a desirable tool for understanding the detailed phenomena involved in precipitation and strengthening, which requires length and timescales far larger than those accessible by first-principles methods. Current interatomic potentials for alloys are not, however, sufficiently accurate for such studies. Here a family of neural-network potentials (NNPs) for the Al-Cu system are presented as a first example of a machine learning potential that can achieve near-first-principles accuracy for many different metallurgically important aspects of this alloy. High-fidelity predictions of intermetallic compounds, elastic constants, dilute solid-solution energetics, precipitate-matrix interfaces, generalized stacking fault energies and surfaces for slip in matrix and precipitates, antisite defect energies, and other quantities, are shown. The NNPs also captures the subtle entropically induced transition between θ′ and θ at temperatures around 600 K. Many comparisons are made with the state-of-the-art angular-dependent potential for Al-Cu, demonstrating the significant quantitative benefit of a machine learning approach. A preliminary kinetic Monte Carlo study shows the NNP to predict the emergence of GP zones in Al-4at\%Cu at T=300 K in agreement with experiments. These studies show that the NNP has significant transferability to defects and properties outside the structures used to train the NNP but also shows some errors highlighting that the use of any interatomic potential requires careful validation in application to specific metallurgical problems of interest.}, - author = {Marchand, Daniel and Jain, Abhinav and Glensk, Albert and Curtin, W. A.}, - date = {2020-10}, - doi = {10.1103/physrevmaterials.4.103601}, - journaltitle = {Physical Review Materials}, - number = {10}, - publisher = {{American Physical Society}}, - title = {Machine Learning for Metallurgy {{I}}. {{A}} Neural-Network Potential for {{Al}}-{{Cu}}}, - volume = {4} -} - -@article{marcolongoSimulatingDiffusionProperties2019, - abstract = {The recently published DeePMD model, based on a deep neural network architecture,1 brings the hope of solving the time-scale issue which often prevents the application of first principle molecular dynamics to physical systems. With this contribution we assess the performance of the DeePMD potential on a real-life application and model diffusion of ions in solid-state electrolytes. We consider as test cases the well known Li10GeP2S12, Li7La3Zr2O12 and Na3Zr2Si2PO12. We develop and test a training protocol suitable for the computation of diffusion coeffcients, which is one of the key properties to be optimized for battery applications, and we find good agreement with previous computations. Our results show that the DeePMD model may be a successful component of a framework to identify novel solid-state electrolytes.}, - archiveprefix = {arXiv}, - author = {Marcolongo, Aris and Binninger, Tobias and Zipoli, Federico and Laino, Teodoro}, - date = {2019}, - eprint = {1910.10090}, - eprinttype = {arxiv}, - shorttitle = {Simulating Diffusion Properties of Solid-State Electrolytes via a Neural Network Potential}, - title = {Simulating Diffusion Properties of Solid-State Electrolytes via a Neural Network Potential: {{Performance}} and Training Scheme} -} - -@article{martelliConnectionLiquidNoncrystalline2020a, - abstract = {The origin of water anomalies hides in an experimentally inaccessible region of the phase diagram known as no-man's land, bounded at low temperature by the domain of stability of amorphous glasses, and at high temperature by the homogeneous nucleation line, below which liquid water loses its metastability. The existence of at least two different forms of glass on one side, i.e., the low-density amorphous (LDA) and the high-density amorphous (HDA) ices, and of one anomalous liquid on the other side, points to a hidden connection between these states, whose understanding has the potential to uncover what happens in no-man's land and shed light on the complex nature of water's behavior. Here, we develop a Neural Network scheme capable of discerning local structures beyond tetrahedrality. Applied over a wide region of the water's phase diagram, we show that the local structures that characterize both LDA and HDA amorphous phases are indeed embedded in the supercooled liquid phase. Remarkably, the rapid increase in the LDA-like population with supercooling occurs in the same temperature and pressure region where thermodynamic fluctuations are maximized, linking these structures with water's anomalies. At the same time, the population of HDA-like environments rapidly increases with pressure, becoming the majority component at high density. Our results show that both LDA and HDA are genuine glasses, and provide a microscopic connection between the non-equilibrium and equilibrium phase diagrams of water.}, - annotation = {WOS:000570952000003}, - author = {Martelli, Fausto and Leoni, Fabio and Sciortino, Francesco and Russo, John}, - date = {2020-09-14}, - doi = {10.1063/5.0018923}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {10}, - pages = {104503}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Connection between Liquid and Non-Crystalline Solid Phases in Water}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{materDeepLearningChemistry2019a, - abstract = {Machine learning enables computers to address problems by learning from data. Deep learning is a type of machine learning that uses a hierarchical recombination of features to extract pertinent information and then learn the patterns represented in the data. Over the last eight years, its abilities have increasingly been applied to a wide variety of chemical challenges, from improving computational chemistry to drug and materials design and even synthesis planning. This review aims to explain the concepts of deep learning to chemists from any background and follows this with an overview of the diverse applications demonstrated in the literature. We hope that this will empower the broader chemical community to engage with this burgeoning field and foster the growing movement of deep learning accelerated chemistry.}, - annotation = {WOS:000473116500006}, - author = {Mater, Adam C. and Coote, Michelle L.}, - date = {2019-06}, - doi = {10.1021/acs.jcim.9b00266}, - issn = {1549-9596}, - journaltitle = {Journal of Chemical Information and Modeling}, - langid = {english}, - location = {{Washington}}, - number = {6}, - pages = {2545--2559}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem Inf. Model.}, - title = {Deep {{Learning}} in {{Chemistry}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {59} -} - -@article{Materialia2021, - abstract = {Large-scale atomistic computer simulations of materials rely on interatomic potentials providing computationally efficient predictions of energy and Newtonian forces. Traditional potentials have served in this capacity for over three decades. Recently, a new class of potentials has emerged, which is based on a radically different philosophy. The new potentials are constructed using machine-learning (ML) methods and a massive reference database generated by quantum-mechanical calculations. While the traditional potentials are derived from physical insights into the nature of chemical bonding, the ML potentials utilize a high-dimensional mathematical regression to interpolate between the reference energies. We review the current status of the interatomic potential field, comparing the strengths and weaknesses of the traditional and ML potentials. A third class of potentials is introduced, in which an ML model is coupled with a physics-based potential to improve the transferability to unknown atomic environments. The discussion is focused on potentials intended for materials science applications. Possible future directions in this field are outlined.}, - archiveprefix = {arXiv}, - arxivid = {2102.06163v2}, - author = {Materialia, Y Mishin - Acta and 2021, undefined}, - date = {2021}, - eprint = {2102.06163v2}, - eprinttype = {arxiv}, - journaltitle = {Elsevier}, - title = {Machine-Learning Interatomic Potentials for Materials Science}, - url = {https://www.sciencedirect.com/science/article/pii/S1359645421003608} -} - -@article{meldgaardMachineLearningEnhanced2018, - abstract = {We show how to speed up global optimization of molecular structures using machine learning methods. To represent the molecular structures, we introduce the auto-bag feature vector that combines (i) a local feature vector for each atom, (ii) an unsupervised clustering of such feature vectors for many atoms across several structures, and (iii) a count for a given structure of how many times each cluster is represented. During subsequent global optimization searches, accumulated structure-energy relations of relaxed structural candidates are used to assign local energies to each atom using supervised learning. Specifically, the local energies follow from assigning energies to each cluster of local feature vectors and demanding the sum of local energies to amount to the structural energies in the least squares sense. The usefulness of the method is demonstrated in basin hopping searches for 19-atom structures described by single- or double-well Lennard-Jones type potentials and for 24-atom carbon structures described by density functional theory. In all cases, utilizing the local energy information derived on-the-fly enhances the rate at which the global minimum energy structure is found. Published by AIP Publishing.}, - annotation = {WOS:000446815600007}, - author = {Meldgaard, Soren A. and Kolsbjerg, Esben L. and Hammer, Bjork}, - date = {2018-10-07}, - doi = {10.1063/1.5048290}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {13}, - pages = {134104}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Machine Learning Enhanced Global Optimization by Clustering Local Environments to Enable Bundled Atomic Energies}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/6}, - urldate = {2021-08-06}, - volume = {149} -} - -@article{meuwlyTransformativeApplicationsMachine2021, - abstract = {Machine learning techniques applied to chemical reactions has a long history. The present contribution discusses applications ranging from small molecule reaction dynamics to platforms for reaction planning. ML-based techniques can be of particular interest for problems which involve both, computation and experiments. For one, Bayesian inference is a powerful approach to include knowledge from experiment in improving computational models. ML-based methods can also be used to handle problems that are formally intractable using conventional approaches, such as exhaustive characterization of state-to-state information in reactive collisions. Finally, the explicit simulation of reactive networks as they occur in combustion has become possible using machine-learned neural network potentials. This review provides an overview of the questions that can and have been addressed using machine learning techniques and an outlook discusses challenges in this diverse and stimulating field. It is concluded that ML applied to chemistry problems as practiced and conceived today has the potential to transform the way with which the field approaches problems involving chemical reactions, both, in research and academic teaching.}, - archiveprefix = {arXiv}, - author = {Meuwly, M.}, - date = {2021-01-10}, - eprint = {2101.03530}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {physics}, - title = {Transformative {{Applications}} of {{Machine Learning}} for {{Chemical Reactions}}}, - url = {http://arxiv.org/abs/2101.03530}, - urldate = {2021-08-11} -} - -@article{miaoLiquidCrystalSi2020, - abstract = {Machine learning force field (ML-FF) has emerged as a potential promising approach to simulate various material phenomena for large systems with ab initio accuracy. However, most ML-FFs have been used to study the phenomena relatively close to the equilibrium ground states. In this work, we have studied a far from equilibrium system of liquid to crystal Si growth using ML-FF. We found that our ML-FF based on ab initio decomposed atomic energy can reproduce all the aspects of ab initio simulated growth, from local energy fluctuations to transition temperatures, to diffusion constant, and growth rates. We have also compared the growth simulation with the Stillinger-Weber classical force field and found significant differences. A procedure is also provided to correct a systematic fitting bias in the ML-FF training process, which exists in all training models, otherwise critical results like transition temperature will be wrong.}, - author = {Miao, Ling and Wang, Lin Wang}, - date = {2020-08}, - doi = {10.1063/5.0011163}, - journaltitle = {Journal of Chemical Physics}, - number = {7}, - publisher = {{American Institute of Physics Inc.}}, - title = {Liquid to Crystal {{Si}} Growth Simulation Using Machine Learning Force Field}, - volume = {153} -} - -@article{mikschStrategiesConstructionMachinelearning2021, - abstract = {Recent advances in machine-learning interatomic potentials have enabled the efficient modeling of complex atomistic systems with an accuracy that is comparable to that of conventional quantum-mechanics based methods. At the same time, the construction of new machine-learning potentials can seem a daunting task, as it involves data-science techniques that are not yet common in chemistry and materials science. Here, we provide a tutorial-style overview of strategies and best practices for the construction of artificial neural network (ANN) potentials. We illustrate the most important aspects of (a) data collection, (b) model selection, (c) training and validation, and (d) testing and refinement of ANN potentials on the basis of practical examples. Current research in the areas of active learning and delta learning are also discussed in the context of ANN potentials. This tutorial review aims at equipping computational chemists and materials scientists with the required background knowledge for ANN potential construction and application, with the intention to accelerate the adoption of the method, so that it can facilitate exciting research that would otherwise be challenging with conventional strategies.}, - annotation = {WOS:000674925500001}, - author = {Miksch, April M. and Morawietz, Tobias and Kaestner, Johannes and Urban, Alexander and Artrith, Nongnuch}, - date = {2021-09}, - doi = {10.1088/2632-2153/abfd96}, - journaltitle = {Machine Learning-Science and Technology}, - langid = {english}, - location = {{Bristol}}, - number = {3}, - pages = {031001}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Mach. Learn.-Sci. Technol.}, - title = {Strategies for the Construction of Machine-Learning Potentials for Accurate and Efficient Atomic-Scale Simulations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {2} -} - -@article{minenkovaGasPhaseSilver2019a, - abstract = {Domain-based local pair natural orbital coupled cluster approach with single, double, and perturbative triple excitations, DLPNO-CCSD(T), has been applied within a framework of a reduced version of the reaction-based Feller-Peterson-Dixon (FPD) scheme to predict gas phase heats of formation and absolute entropies of silver inorganic and organometallic compounds. First, we evaluated all existing experimental data currently limited by thermodynamic functions of 10 silver substances (AgH, AgF, AgBr, AgI, Ag-2, Ag2S, Ag2Se, Ag2Te, AgCN, AgPO2). The mean average deviation between computed and experimental heats of formation was found to be 1.9 kcal/mol. Notably, all predicted heats of formation turned out to be within the error bounds of their experimental counterparts. Second, we predicted heats of formation and entropies for additional 90 silver species with no experimental data available, substantially enriching silver thermochemistry. Combination of gas phase heats of formation Delta H-f and entropies S degrees of AgNO2, AgSCN, Ag2SO4, and Ag2SeO4 obtained in this work, with respective solid-state information, resulted in accurate sublimation thermochemistry of these compounds. Complementation of predicted Delta H-f with heats of formation of some neutrals and positive ions produced 33 silver bond strengths of high reliability. Obtained thermochemical data are promising for developing the concepts of silver chemistry. In addition, derived heats of formation and bond dissociation enthalpies, due to their high diversity, are found to be relevant for testing and training of computational chemistry methods.}, - annotation = {WOS:000472241400029}, - author = {Minenkova, Irina and Slizney, Valery V. and Cavallo, Luigi and Minenkov, Yury}, - date = {2019-06-17}, - doi = {10.1021/acs.inorgchem.9b00556}, - issn = {0020-1669}, - journaltitle = {Inorganic Chemistry}, - langid = {english}, - location = {{Washington}}, - number = {12}, - pages = {7873--7885}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {Inorg. Chem.}, - title = {Gas {{Phase Silver Thermochemistry}} from {{First Principles}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {58} -} - -@article{Mirhosseini2021, - abstract = {The performance of machine learning interatomic potentials relies on the quality of the training dataset. In this work, we present an approach for generating diverse and representative training data points which initiates with ab initio calculations for bulk structures. The data generation and potential construction further proceed side-by-side in a cyclic process of training the neural network and crystal structure prediction based on the developed interatomic potentials. All steps of the data generation and potential development are performed with minimal human intervention. We show the reliability of our approach by assessing the performance of neural network potentials developed for two inorganic systems.}, - archiveprefix = {arXiv}, - arxivid = {2102.04085v1}, - author = {Mirhosseini, H and Tahmasbi, H and …, SR Kuchana - Computational Materials and 2021, undefined}, - date = {2021}, - eprint = {2102.04085v1}, - eprinttype = {arxiv}, - journaltitle = {Elsevier}, - title = {An Automated Approach for Developing Neural Network Interatomic Potentials with {{FLAME}}}, - url = {https://www.sciencedirect.com/science/article/pii/S0927025621002949} -} - -@article{moradzadehMolecularDynamicsProperties2019a, - abstract = {Molecular dynamics (MD) simulation is a popularly used computational tool to compute microscopic and macroscopic properties of a variety of systems including liquids, solids, biological systems, etc. To determine properties of atomic systems to a good level of accuracy with minimal noise or fluctuation, MD simulations are performed over a long time ranging from a few nanoseconds to several tens to hundreds of nanoseconds depending on the system and the properties of interest. In this study, by considering simple liquids, we explore the feasibility of significantly reducing the MD simulation time to compute various properties of monatomic systems such as the structure, pressure, and isothermal compressibility. To do so, extensive MD simulations are performed on 12 000 distinct Lennard-Jones systems at various thermodynamic states. Then, a deep denoising autoencoder network is trained to take the radial distribution function (RDF) from a single snapshot of a Lennard-Jones liquid to compute the mean, temporally averaged RDF. We show that the method is successful in the prediction of RDF and other properties such as the pressure and isothermal compressibility that can be computed based on the RDF not only for Lennard-Jones liquids at various thermodynamic states but also for various simple liquids described by exponential, Yukawa, and inverse-power-law pair potentials.}, - annotation = {WOS:000503919300005}, - author = {Moradzadeh, Alireza and Aluru, N. R.}, - date = {2019-12-19}, - doi = {10.1021/acs.jpclett.9b02820}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {24}, - pages = {7568--7576}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - shorttitle = {Molecular {{Dynamics Properties}} without the {{Full Trajectory}}}, - title = {Molecular {{Dynamics Properties}} without the {{Full Trajectory}}: {{A Denoising Autoencoder Network}} for {{Properties}} of {{Simple Liquids}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {10} -} - -@article{moralesDeepLearningGravitationalWave2021, - abstract = {In this work, we apply Convolutional Neural Networks (CNNs) to detect gravitational wave (GW) signals of compact binary coalescences, using single-interferometer data from real LIGO detectors. Here, we adopted a resampling white-box approach to advance towards a statistical understanding of uncertainties intrinsic to CNNs in GW data analysis. We used Morlet wavelets to convert strain time series to time-frequency images. Moreover, we only worked with data of non-Gaussian noise and hardware injections, removing freedom to set signal-to-noise ratio (SNR) values in GW templates by hand, in order to reproduce more realistic experimental conditions. After hyperparameter adjustments, we found that resampling through repeated k-fold cross-validation smooths the stochasticity of mini-batch stochastic gradient descent present in accuracy perturbations by a factor of 3.6. CNNs are quite precise to detect noise, 0.952 for H1 data and 0.932 for L1 data; but, not sensitive enough to recall GW signals, 0.858 for H1 data and 0.768 for L1 data—although recall values are dependent on expected SNR. Our predictions are transparently understood by exploring tthe distribution of probabilistic scores outputted by the softmax layer, and they are strengthened by a receiving operating characteristic analysis and a paired-sample t-test to compare with a random classifier.}, - author = {Morales, Manuel D. and Antelis, Javier M. and Moreno, Claudia and Nesterov, Alexander I.}, - date = {2021-05-03}, - doi = {10/gmgfm6}, - issn = {1424-8220}, - journaltitle = {Sensors}, - langid = {english}, - number = {9}, - pages = {3174}, - shortjournal = {Sensors}, - shorttitle = {Deep {{Learning}} for {{Gravitational}}-{{Wave Data Analysis}}}, - title = {Deep {{Learning}} for {{Gravitational}}-{{Wave Data Analysis}}: {{A Resampling White}}-{{Box Approach}}}, - url = {https://www.mdpi.com/1424-8220/21/9/3174}, - urldate = {2021-08-11}, - volume = {21} -} - -@article{morawietzMachineLearningacceleratedQuantum2021, - abstract = {Atomistic simulations have become an invaluable tool for industrial applications ranging from the optimization of protein-ligand interactions for drug discovery to the design of new materials for energy applications. Here we review recent advances in the use of machine learning (ML) methods for accelerated simulations based on a quantum mechanical (QM) description of the system. We show how recent progress in ML methods has dramatically extended the applicability range of conventional QM-based simulations, allowing to calculate industrially relevant properties with enhanced accuracy, at reduced computational cost, and for length and time scales that would have otherwise not been accessible. We illustrate the benefits of ML-accelerated atomistic simulations for industrial R\&D processes by showcasing relevant applications from two very different areas, drug discovery (pharmaceuticals) and energy materials. Writing from the perspective of both a molecular and a materials modeling scientist, this review aims to provide a unified picture of the impact of ML-accelerated atomistic simulations on the pharmaceutical, chemical, and materials industries and gives an outlook on the exciting opportunities that could emerge in the future.}, - annotation = {WOS:000577932000001}, - author = {Morawietz, Tobias and Artrith, Nongnuch}, - date = {2021-04}, - doi = {10.1007/s10822-020-00346-6}, - issn = {0920-654X}, - journaltitle = {Journal of Computer-Aided Molecular Design}, - langid = {english}, - location = {{Dordrecht}}, - number = {4}, - pages = {557--586}, - publisher = {{Springer}}, - shortjournal = {J. Comput.-Aided Mol. Des.}, - title = {Machine Learning-Accelerated Quantum Mechanics-Based Atomistic Simulations for Industrial Applications}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {35} -} - -@article{moTransferLearningPotential2020, - abstract = {This letter proposes a transfer learning (TL) method to generate neural network (NN) database to model doping and alloy. By leveraging the valuable potential energy surface (PES) information already available in source system and similarities between source and target systems, the proposed TL successfully reduces computational cost by several orders of magnitude, while keeping ab-initio level high accuracy. We show that it is generally applicable to model p-type, n-type, and alloy atomic substitutions.}, - author = {Mo, Pinghui and Shi, Mengchao and Yao, Wenze and Liu, Jie}, - date = {2020-04}, - doi = {10/gg2bfc}, - issn = {0741-3106, 1558-0563}, - journaltitle = {IEEE Electron Device Letters}, - langid = {english}, - number = {4}, - pages = {633--636}, - shortjournal = {IEEE Electron Device Lett.}, - title = {Transfer {{Learning}} of {{Potential Energy Surfaces}} for {{Efficient Atomistic Modeling}} of {{Doping}} and {{Alloy}}}, - url = {https://ieeexplore.ieee.org/document/8985427/}, - urldate = {2021-08-10}, - volume = {41} -} - -@article{moussaAssessmentLocalizedRandomized2019a, - abstract = {As electronic structure simulations continue to grow in size, the system-size scaling of computational costs increases in importance relative to cost prefactors. Presently, linear-scaling costs for three-dimensional systems are only attained by localized or randomized algorithms that have large cost prefactors in the difficult regime of low-temperature metals. Using large copper clusters in a minimal-basis semiempirical model as our reference system, we study the costs of these algorithms relative to a conventional cubic-scaling algorithm using matrix diagonalization and a recent quadratic-scaling algorithm using sparse matrix factorization and rational function approximation. The linear-scaling algorithms are competitive at the high temperatures relevant for warm dense matter, but their cost prefactors are prohibitive near ambient temperatures. To further reduce costs, we consider hybridized algorithms that combine localized and randomized algorithms. While simple hybridized algorithms do not improve performance, more sophisticated algorithms using recent concepts from structured linear algebra show promising initial performance results on a simple-cubic orthogonal tight-binding model.}, - annotation = {WOS:000604917000001}, - author = {Moussa, Jonathan E. and Baczewski, Andrew D.}, - date = {2019-09}, - doi = {10.1088/2516-1075/ab2022}, - issn = {2516-1075}, - journaltitle = {Electronic Structure}, - langid = {english}, - location = {{Bristol}}, - number = {3}, - pages = {033001}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Electron. Struct.}, - title = {Assessment of Localized and Randomized Algorithms for Electronic Structure}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {1} -} - -@article{mrazekDynamicControlLight2019, - abstract = {The article is focused on the issue of interval on a light signaling device. Light signaling devices operate on different systems by means of which they are controlled. The control problem is a very static setting that does not respond to real-time traffic. Important variables for dynamic real-time control are traffic density in a selected area along with average speed. These variables are interdependent and can be based on dynamic traffic control. Dynamic traffic control ensures smoother traffic through major turns. At the same time, the number of harmful CO2 emitted from the means of transport should be reduced to the air. When used in low operation, power consumption should be reduced.}, - author = {Mrazek, Jan and Mrazkova, Lucia Duricova and Hromada, Martin and Reznickova, Jana}, - date = {2019}, - doi = {10/gmgfts}, - editor = {Mastorakis, N. and Mladenov, V. and Bulucea, A.}, - issn = {2261-236X}, - journaltitle = {MATEC Web of Conferences}, - langid = {english}, - pages = {03014}, - shortjournal = {MATEC Web Conf.}, - title = {The {{Dynamic Control}} of the {{Light Signalling Device}} in {{Real}}-{{Time}}}, - url = {https://www.matec-conferences.org/10.1051/matecconf/201929203014}, - urldate = {2021-08-11}, - volume = {292} -} - -@inproceedings{mrazekTrafficControlTraffic2019, - abstract = {The article is focused on dynamic driving in road transport. Current systems on the market are referred to as dynamic but have their limitations. Traffic management is an important aspect to ensure the proper functioning and safety of transport as an element of critical infrastructure. Ensuring the prevention of possible threats can minimize the possibility of threatening other elements of critical infrastructure. The proposed method deals with dynamic driving in road transport. Dynamically controlled traffic lights should ensure greater smoothness, safety and reduce the number of harmful substances emitting vehicles.}, - annotation = {WOS:000647641800004}, - author = {Mrazek, Jan and Mrazkova, Lucia Duricova and Hromada, Martin}, - booktitle = {2019 3rd {{European Conference}} on {{Electrical Engineering}} and {{Computer Science}} (Eecs 2019)}, - date = {2019}, - doi = {10.1109/EECS49779.2019.00017}, - isbn = {978-1-72816-109-9}, - langid = {english}, - location = {{Los Alamitos}}, - pages = {19--21}, - publisher = {{Ieee Computer Soc}}, - title = {Traffic {{Control Through Traffic Density}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/6}, - urldate = {2021-08-06} -} - -@article{muellerMachineLearningInteratomic2020a, - abstract = {The use of supervised machine learning to develop fast and accurate interatomic potential models is transforming molecular and materials research by greatly accelerating atomic-scale simulations with little loss of accuracy. Three years ago, Jorg Behler published a perspective in this journal providing an overview of some of the leading methods in this field. In this perspective, we provide an updated discussion of recent developments, emerging trends, and promising areas for future research in this field. We include in this discussion an overview of three emerging approaches to developing machine-learned interatomic potential models that have not been extensively discussed in existing reviews: moment tensor potentials, message-passing networks, and symbolic regression. Published under license by AIP Publishing.}, - annotation = {WOS:000531243200002}, - author = {Mueller, Tim and Hernandez, Alberto and Wang, Chuhong}, - date = {2020-02-07}, - doi = {10.1063/1.5126336}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {5}, - pages = {050902}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Machine Learning for Interatomic Potential Models}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {152} -} - -@article{mujalSupervisedLearningFew2020, - abstract = {Supervised machine learning is emerging as a powerful computational tool to predict the properties of complex quantum systems at a limited computational cost. In this article, we quantify how accurately deep neural networks can learn the properties of disordered quantum systems as a function of the system size. We implement a scalable convolutional network that can address arbitrary system sizes. This network is compared with a recently introduced extensive convolutional architecture [K. Mills et al., Chem. Sci. 10, 4129 (2019)] and with conventional dense networks with all-to-all connectivity. The networks are trained to predict the exact ground-state energies of various disordered systems, namely a continuous-space single-particle Hamiltonian for cold-atoms in speckle disorder, and different setups of a quantum Ising chain with random couplings, including one with only short-range interactions and one augmented with a long-range term. In all testbeds we consider, the scalable network retains high accuracy as the system size increases. Furthermore, we demonstrate that the network scalability enables a transfer-learning protocol, whereby a pre-training performed on small systems drastically accelerates the learning of large-system properties, allowing reaching high accuracy with small training sets. In fact, with the scalable network one can even extrapolate to sizes larger than those included in the training set, accurately reproducing the results of state-of-the-art quantum Monte Carlo simulations.}, - archiveprefix = {arXiv}, - arxivid = {2005.14290v1}, - author = {Mujal, P and Miguel, À Martínez and Polls, A}, - date = {2020-10-08}, - eprint = {2005.14290v1}, - eprinttype = {arxiv}, - journaltitle = {scipost.org}, - title = {Supervised Learning of Few Dirty Bosons with Variable Particle Number}, - url = {https://www.scipost.org/SciPostPhys.10.3.073/pdf} -} - -@article{musilMachineLearningAtomic2019, - abstract = {Statistical learning algorithms are finding more and more applications in science and technology. Atomic-scale modeling is no exception, with machine learning becoming commonplace as a tool to predict energy, forces and properties of molecules and condensed-phase systems. This short review summarizes recent progress in the field, focusing in particular on the problem of representing an atomic configuration in a mathematically robust and computationally efficient way. We also discuss some of the regression algorithms that have been used to construct surrogate models of atomic-scale properties. We then show examples of how the optimization of the machine-learning models can both incorporate and reveal insights onto the physical phenomena that underlie structure-property relations.}, - annotation = {WOS:000504762100002}, - author = {Musil, Felix and Ceriotti, Michele}, - date = {2019-12}, - doi = {10.2533/chimia.2019.972}, - issn = {0009-4293}, - journaltitle = {Chimia}, - langid = {english}, - location = {{Bern}}, - number = {12}, - pages = {972--982}, - publisher = {{Swiss Chemical Soc}}, - shortjournal = {Chimia}, - title = {Machine {{Learning}} at the {{Atomic Scale}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {73} -} - -@article{nagornovNonempiricalWeightedLangevin2019, - abstract = {Recently a non-empirical stochastic walker algorithm has been developed to search for the minimum-energy escape paths (MEP) from the minima of the potential surface [J. Phys. Soc. Jpn. 87, 063801 (2018)]. This algorithm is novel in that it tracks the MEP monotonically and does not use the whole Hessian matrix but only gradient and Laplacian of the potential. In this work, we implement an MPI-parallelized version of this algorithm in a simple way. We also explore efficient ways to reduce the number of walkers required for the accurate tracking of the MEP and generate initial positions automatically. We apply the whole scheme to the Lennard-Jones argon cluster with 7-38 atoms to demonstrate the successful tracking of the reaction paths. This achievement paves the path to non-empirical simulation of rare reactions without coarse-graining or artificial potential.}, - archiveprefix = {arXiv}, - author = {Nagornov, Yuri S. and Akashi, Ryosuke}, - date = {2019-08}, - doi = {10.1016/j.physa.2019.121481}, - eprint = {1812.06581}, - eprinttype = {arxiv}, - issn = {03784371}, - journaltitle = {Physica A: Statistical Mechanics and its Applications}, - langid = {english}, - note = {Comment: 30 pages (single column), 13 figures}, - pages = {121481}, - shortjournal = {Physica A: Statistical Mechanics and its Applications}, - shorttitle = {Non-Empirical Weighted {{Langevin}} Mechanics for the Potential Escape Problem}, - title = {Non-Empirical Weighted {{Langevin}} Mechanics for the Potential Escape Problem: Parallel Algorithm and Application to the {{Argon}} Clusters}, - url = {http://arxiv.org/abs/1812.06581}, - urldate = {2021-08-11}, - volume = {528} -} - -@article{niblettLearningIntermolecularForces2021, - abstract = {By adopting a perspective informed by contemporary liquid state theory, we consider how to train an artificial neural network potential to describe inhomogeneous, disordered systems. Specifically, we find that neural network potentials based on local representations of atomic environments are capable of describing bulk liquid properties, but can fail at liquid-vapor interfaces. This failure results from the unbalanced attractions that build up in the presence of broken translation symmetry from long-ranged interactions, but cancel in the translational invariant bulk. By incorporating explicit models of the slowly-varying long-ranged interactions and training neural networks only on the short ranged components, we can arrive at potentials that robustly recover interfacial properties. We find that local neural network models can sometimes approximate a local molecular field potential to correct for the truncated interactions, but this behavior is variable and hard to learn. Generally, we find that models with explicit electrostatics are easier to train and have higher accuracy. We demonstrate this perspective in a simple model of an asymmetric dipolar fluid where the exact long-ranged interaction is known, and in an ab initio water model where it is approximated.}, - archiveprefix = {arXiv}, - author = {Niblett, Samuel P. and Galib, Mirza and Limmer, David T.}, - date = {2021-07-13}, - eprint = {2107.06208}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 11 pages, 8 Figures. Comments welcome}, - primaryclass = {cond-mat, physics:physics}, - title = {Learning Intermolecular Forces at Liquid-Vapor Interfaces}, - url = {http://arxiv.org/abs/2107.06208}, - urldate = {2021-08-11} -} - -@article{nigamRecursiveEvaluationIterative2020a, - abstract = {Mapping an atomistic configuration to a symmetrized N-point correlation of a field associated with the atomic positions (e.g., an atomic density) has emerged as an elegant and effective solution to represent structures as the input of machine-learning algorithms. While it has become clear that low-order density correlations do not provide a complete representation of an atomic environment, the exponential increase in the number of possible N-body invariants makes it difficult to design a concise and effective representation. We discuss how to exploit recursion relations between equivariant features of different order (generalizations of N-body invariants that provide a complete representation of the symmetries of improper rotations) to compute high-order terms efficiently. In combination with the automatic selection of the most expressive combination of features at each order, this approach provides a conceptual and practical framework to generate systematically improvable, symmetry adapted representations for atomistic machine learning.}, - annotation = {WOS:000575139300001}, - author = {Nigam, Jigyasa and Pozdnyakov, Sergey and Ceriotti, Michele}, - date = {2020-09-28}, - doi = {10.1063/5.0021116}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {12}, - pages = {121101}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Recursive Evaluation and Iterative Contraction of {{N}}-Body Equivariant Features}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{nikolovQuantumaccurateMagnetoelasticPredictions2021, - abstract = {A data-driven framework is presented for building magneto-elastic machine-learning interatomic potentials (ML-IAPs) for large-scale spin-lattice dynamics simulations. The magneto-elastic ML-IAPs are constructed by coupling a collective atomic spin model with an ML-IAP. Together they represent a potential energy surface from which the mechanical forces on the atoms and the precession dynamics of the atomic spins are computed. Both the atomic spin model and the ML-IAP are parametrized on data from first-principles calculations. We demonstrate the efficacy of our data-driven framework across magneto-structural phase transitions by generating a magneto-elastic ML-IAP for \{\textbackslash alpha\}-iron. The combined potential energy surface yields excellent agreement with first-principles magneto-elastic calculations and quantitative predictions of diverse materials properties including bulk modulus, magnetization, and specific heat across the ferromagnetic-paramagnetic phase transition.}, - archiveprefix = {arXiv}, - author = {Nikolov, Svetoslav and Wood, Mitchell A. and Cangi, Attila and Maillet, Jean-Bernard and Marinica, Mihai-Cosmin and Thompson, Aidan P. and Desjarlais, Michael P. and Tranchida, Julien}, - date = {2021-02-02}, - eprint = {2101.07332}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 15 pages, 6 figures, 2 tables}, - primaryclass = {cond-mat}, - title = {Quantum-Accurate Magneto-Elastic Predictions with Classical Spin-Lattice Dynamics}, - url = {http://arxiv.org/abs/2101.07332}, - urldate = {2021-08-11} -} - -@article{niuInitioPhaseDiagram2020a, - abstract = {Elemental gallium possesses several intriguing properties, such as a low melting point, a density anomaly and an electronic structure in which covalent and metallic features coexist. In order to simulate this complex system, we construct an ab initio quality interaction potential by training a neural network on a set of density functional theory calculations performed on configurations generated in multithermal-multibaric simulations. Here we show that the relative equilibrium between liquid gallium, alpha-Ga, beta-Ga, and Ga-II is well described. The resulting phase diagram is in agreement with the experimental findings. The local structure of liquid gallium and its nucleation into alpha-Ga and beta-Ga are studied. We find that the formation of metastable beta-Ga is kinetically favored over the thermodinamically stable alpha-Ga. Finally, we provide insight into the experimental observations of extreme undercooling of liquid Ga.}, - annotation = {WOS:000538030400010}, - author = {Niu, Haiyang and Bonati, Luigi and Piaggi, Pablo M. and Parrinello, Michele}, - date = {2020-05-27}, - doi = {10.1038/s41467-020-16372-9}, - issn = {2041-1723}, - journaltitle = {Nature Communications}, - langid = {english}, - location = {{London}}, - number = {1}, - pages = {2654}, - publisher = {{Nature Publishing Group}}, - shortjournal = {Nat. Commun.}, - title = {Ab Initio Phase Diagram and Nucleation of Gallium}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{novikovMLIPPackageMoment2021, - abstract = {The subject of this paper is the technology (the 'how') of constructing machine-learning interatomic potentials, rather than science (the 'what' and 'why') of atomistic simulations using machine-learning potentials. Namely, we illustrate how to construct moment tensor potentials using active learning as implemented in the MLIP package, focusing on the efficient ways to automatically sample configurations for the training set, how expanding the training set changes the error of predictions, how to set up ab initio calculations in a cost-effective manner, etc. The MLIP package (short for Machine-Learning Interatomic Potentials) is available at https://mlip.skoltech.ru/download/.}, - annotation = {WOS:000660868300001}, - author = {Novikov, Ivan S. and Gubaev, Konstantin and Podryabinkin, Evgeny and Shapeev, Alexander}, - date = {2021-06}, - doi = {10.1088/2632-2153/abc9fe}, - journaltitle = {Machine Learning-Science and Technology}, - langid = {english}, - location = {{Bristol}}, - number = {2}, - pages = {025002}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Mach. Learn.-Sci. Technol.}, - shorttitle = {The {{MLIP}} Package}, - title = {The {{MLIP}} Package: Moment Tensor Potentials with {{MPI}} and Active Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {2} -} - -@inproceedings{oehmckeModelingH2ORutileTiO22020, - abstract = {Successful and cost-effective water splitting could be one of the most interesting energy sources of the future. The modification of the transition metal oxide rutile-TiO2 as photocatalyst can lead to improved performance in the water splitting process. For that purpose, an accurate description of the interaction potential of a water molecule and the rutile-TiO2(110) surface in the ground and electronically excited state after photoexcitation is crucial. The electronic Schrodinger equation for the states involved is solved pointwise for different nuclear configurations within the Born-Oppenheimer approximation, and accurate fits to these energy points are required to obtain an analytic expression for the potential energy surface. This is too computationally expensive for fine-grained surface calculations of quantum chemical models. In this paper, we propose to use state-of-the-art deep learning techniques to provide accurate fits for this problem. Namely, we employ a fully connected variant of ResNet and DenseNet with heavy regularization (L2, RReLU, Dropout, and BatchNormalization). Previous literature applied neural network approaches before, but with unsatisfactory accuracy. In an experimental evaluation we show that the root mean squared error (RMSE) can be 6.8 times lower for the exited state and 12.7 times lower for the ground state compared to former approaches.}, - annotation = {WOS:000626021405049}, - author = {Oehmcke, Stefan and Teusch, Thomas and Petersen, Thorben and Kluener, Thorsten and Kramer, Oliver}, - booktitle = {2020 {{International Joint Conference}} on {{Neural Networks}} (Ijcnn)}, - date = {2020}, - isbn = {978-1-72816-926-2}, - issn = {2161-4393}, - langid = {english}, - location = {{New York}}, - publisher = {{Ieee}}, - title = {Modeling {{H2O}}/{{Rutile}}-{{TiO2}}(110) {{Potential Energy Surfaces}} with {{Deep Networks}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06} -} - -@article{orupatturCatalyticMaterialsChemistry2020a, - abstract = {First principles-based molecular modelling plays a crucial role in the development of novel catalytic materials and in the investigation of catalytic chemical reactions. However, the computational cost and/or the accuracy of these models remains a bottleneck in carrying out these simulations for complex or large scale systems, as in the case of catalysis. Over the past two decades, machine learning (ML) has made an impact in the field of computational catalysis. Modern-day researchers have started using machine learning-based data-driven techniques to overcome the limitations of these molecular simulations. In this review, we summarize the recent progress in the utilization of ML algorithms to assist molecular simulations, followed by its applications in the field of catalysis. Furthermore, we provide our perspective on promising avenues for research in the future regarding the incorporation of ML in molecular simulations in catalysis.}, - annotation = {WOS:000509344200006}, - author = {Orupattur, Nilesh Varadan and Mushrif, Samir H. and Prasad, Vinay}, - date = {2020-03}, - doi = {10.1016/j.commatsci.2019.109474}, - issn = {0927-0256}, - journaltitle = {Computational Materials Science}, - langid = {english}, - location = {{Amsterdam}}, - pages = {109474}, - publisher = {{Elsevier}}, - shortjournal = {Comput. Mater. Sci.}, - title = {Catalytic Materials and Chemistry Development Using a Synergistic Combination of Machine Learning and Ab Initio Methods}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {174} -} - -@article{paleicoBinHashMethod2021, - abstract = {In recent years the development of machine learning potentials (MLPs) has become a very active field of research. Numerous approaches have been proposed, which allow one to perform extended simulations of large systems at a small fraction of the computational costs of electronic structure calculations. The key to the success of modern MLPs is the close-to first principles quality description of the atomic interactions. This accuracy is reached by using very flexible functional forms in combination with high-level reference data from electronic structure calculations. These data sets can include up to hundreds of thousands of structures covering millions of atomic environments to ensure that all relevant features of the potential energy surface are well represented. The handling of such large data sets is nowadays becoming one of the main challenges in the construction of MLPs. In this paper we present a method, the bin-and-hash (BAH) algorithm, to overcome this problem by enabling the efficient identification and comparison of large numbers of multidimensional vectors. Such vectors emerge in multiple contexts in the construction of MLPs. Examples are the comparison of local atomic environments to identify and avoid unnecessary redundant information in the reference data sets that is costly in terms of both the electronic structure calculations as well as the training process, the assessment of the quality of the descriptors used as structural fingerprints in many types of MLPs, and the detection of possibly unreliable data points. The BAH algorithm is illustrated for the example of high-dimensional neural network potentials using atom-centered symmetry functions for the geometrical description of the atomic environments, but the method is general and can be combined with any current type of MLP.}, - annotation = {WOS:000660870100001}, - author = {Paleico, Martin Leandro and Behler, Joerg}, - date = {2021-09}, - doi = {10.1088/2632-2153/abe663}, - journaltitle = {Machine Learning-Science and Technology}, - langid = {english}, - location = {{Bristol}}, - number = {3}, - pages = {037001}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Mach. Learn.-Sci. Technol.}, - title = {A Bin and Hash Method for Analyzing Reference Data and Descriptors in Machine Learning Potentials}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {2} -} - -@article{Pan2021, - abstract = {Despite recent advances in the development of machine learning potentials (MLPs) for biomolecular simulations, there has been limited effort in developing stable and accurate MLPs for enzymatic reactions. Here, we report a protocol for performing machine learning assisted free energy simulation of solution-phase and enzyme reactions at an ab initio quantum mechanical and molecular mechanical (ai-QM/MM) level of accuracy. Within our protocol, the MLP is built to reproduce the ai-QM/MM energy as well as forces on both QM (reactive) and MM (solvent/enzyme) atoms. As an alternative strategy, a delta machine learning potential (DMLP) is trained to reproduce the differences between ai-QM/MM and semi-empirical (se) QM/MM energy and forces. To account for the effect of the condensed-phase environment in both MLP and DMLP, the DeePMD representation of a molecular system is extended to incorporate external electrostatic potential and field on each QM atom. Using the Menshutkin and chorismate mutase reactions as examples , we show that the developed MLP and DMLP reproduce the ai-QM/MM energy and forces with an error on average less than 1.0 kcal/mol and 1.0 kcal/mol/Å for representative configurations along the reaction pathway. For both reactions, MLP/DMLP-based simulations yielded free energy profiles that differed by less than 1.0 kcal/mol from the reference ai-QM/MM results, but only at a fractional computational cost.}, - author = {Pan, X and Van, R and Epifanovsky, E and Ho, J and Huang, J and Pu, J}, - date = {2021}, - title = {Machine Learning Assisted Free Energy Simulation of {{Solution}}–{{Phase}} and Enzyme Reactions}, - url = {https://chemrxiv.org/engage/chemrxiv/article-details/60c73cc9bdbb89b390a37b7d} -} - -@article{panDFTAccurateMachine2020a, - abstract = {Molten eutectic salts consisting of ZnCl2 and other alkali chlorides are promising thermal storage and heat transfer fluid materials in the next generation concentrated solar thermal power. To go deep into the thermal and transport properties for a high order mixture, the microstructure information, as well as thermodynamics properties of individual components, have to be identified first. This work develops interatomic potentials of molten ZnCl2 based on neural-network machine learning approach for the first time. The machine learning potential is trained by fitting to the energies and forces of liquid structures ab initio molecular dynamics calculations. The developed machine learning potential is validated by comparing partial radial distribution functions, coordination numbers, and partial structure factors with AIMD and PIM potential. The machine learning potential yields a more precise description of the microstructures than the PIM potential which suffers from the analytical form. Furthermore, structural and thermophysical evolution with temperature are studied and the results are in good agreement with experimental values. The efficient machine learning potential with DFT accuracy from our study will provide a promising scheme for accurate molecular simulations of structures and dynamics of molten ZnCl2 mixtures.}, - annotation = {WOS:000579155500014}, - author = {Pan, Gechuanqi and Chen, Pin and Yan, Hui and Lu, Yutong}, - date = {2020-12}, - doi = {10.1016/j.commatsci.2020.109955}, - issn = {0927-0256}, - journaltitle = {Computational Materials Science}, - langid = {english}, - location = {{Amsterdam}}, - pages = {109955}, - publisher = {{Elsevier}}, - shortjournal = {Comput. Mater. Sci.}, - shorttitle = {A {{DFT}} Accurate Machine Learning Description of Molten {{ZnCl2}} and Its Mixtures}, - title = {A {{DFT}} Accurate Machine Learning Description of Molten {{ZnCl2}} and Its Mixtures: 1. {{Potential}} Development and Properties Prediction of Molten {{ZnCl2}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {185} -} - -@article{panDFTAccurateMachine2021a, - abstract = {ZnCl2-NaCl-KCl ternary salts are promising thermal storage and heat transfer fluid materials with a freezing point below 250 degrees C, thermal stability up to 800 degrees C, and other favorable properties that fit the use in the next generation concentrated solar thermal power. This work for the first time developed a machine learning-based interatomic potential for ZnCl2-NaCl-KCl ternary salt (0.6:0.2:0.2 in mole fraction) on the basis of energies and forces estimated by ab initio molecular dynamics calculations. The proposed machine learning potential was validated with the obtained partial radial distribution functions and the coordination numbers with the AIMD. The structural and thermophysical evolutions with temperature over the entire operating temperature range were documented. Adding Na+ and K+ ions deteriorated the network by corner-sharing and edge-sharing ZnCl4 tetrahedra, and apparently affected self-diffusion coefficient, thermal conductivity, and viscosity of the melt. The calculated thermophysical properties agreed with experimental data. A negative temperature dependence of thermal conductivity was noted and discussed. Based on the experimental data, viscosity data by Li et al. and those of this work, yielded reliable experimental values in the Vogel-Tamman-Fulcher form.}, - annotation = {WOS:000600373900010}, - author = {Pan, Gechuanqi and Ding, Jing and Du, Yunfei and Lee, Duu-Jong and Lu, Yutong}, - date = {2021-02-01}, - doi = {10.1016/j.commatsci.2020.110055}, - issn = {0927-0256}, - journaltitle = {Computational Materials Science}, - langid = {english}, - location = {{Amsterdam}}, - pages = {110055}, - publisher = {{Elsevier}}, - shortjournal = {Comput. Mater. Sci.}, - shorttitle = {A {{DFT}} Accurate Machine Learning Description of Molten {{ZnCl2}} and Its Mixtures}, - title = {A {{DFT}} Accurate Machine Learning Description of Molten {{ZnCl2}} and Its Mixtures: 2. {{Potential}} Development and Properties Prediction of {{ZnCl2}}-{{NaCl}}-{{KCl}} Ternary Salt for {{CSP}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {187} -} - -@article{parkAccurateScalableGraph2021, - abstract = {Recently, machine learning (ML) has been used to address the computational cost that has been limiting ab initio molecular dynamics (AIMD). Here, we present GNNFF, a graph neural network framework to directly predict atomic forces from automatically extracted features of the local atomic environment that are translationally-invariant, but rotationally-covariant to the coordinate of the atoms. We demonstrate that GNNFF not only achieves high performance in terms of force prediction accuracy and computational speed on various materials systems, but also accurately predicts the forces of a large MD system after being trained on forces obtained from a smaller system. Finally, we use our framework to perform an MD simulation of Li7P3S11, a superionic conductor, and show that resulting Li diffusion coefficient is within 14\% of that obtained directly from AIMD. The high performance exhibited by GNNFF can be easily generalized to study atomistic level dynamics of other material systems.}, - annotation = {WOS:000658724900002}, - author = {Park, Cheol Woo and Kornbluth, Mordechai and Vandermause, Jonathan and Wolverton, Chris and Kozinsky, Boris and Mailoa, Jonathan P.}, - date = {2021-05-21}, - doi = {10.1038/s41524-021-00543-3}, - journaltitle = {Npj Computational Materials}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - pages = {73}, - publisher = {{Nature Research}}, - shortjournal = {npj Comput. Mater.}, - title = {Accurate and Scalable Graph Neural Network Force Field and Molecular Dynamics with Direct Force Architecture}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {7} -} - -@article{parkAccurateScalableMultielement, - abstract = {Recently, machine learning (ML) has been used to address the computational cost that has been limiting ab initio molecular dynamics (AIMD). Here, we present GNNFF, a graph neural network framework to directly predict atomic forces from automatically extracted features of the local atomic environment that are translationally-invariant, but rotationally-covariant to the coordinate of the atoms. We demonstrate that GNNFF not only achieves high performance in terms of force prediction accuracy and computational speed on various materials systems, but also accurately predicts the forces of a large MD system after being trained on forces obtained from a smaller system. Finally, we use our framework to perform an MD simulation of Li7P3S11, a superionic conductor, and show that resulting Li diffusion coefficient is within 14\% of that obtained directly from AIMD. The high performance exhibited by GNNFF can be easily generalized to study atomistic level dynamics of other material systems.}, - author = {Park, Cheol Woo and Kornbluth, Mordechai and Vandermause, Jonathan and Wolverton, Chris and Mailoa, Jonathan P}, - langid = {english}, - pages = {33}, - title = {Accurate and Scalable Multi-Element Graph Neural Network Force Field and Molecular Dynamics with Direct Force Architecture}, - url = {https://arxiv.org/ftp/arxiv/papers/2007/2007.14444.pdf} -} - -@article{peigneyFourierbasedMachineLearning2021a, - abstract = {The generic problem in supervised machine learning is to learn a function f from a collection of samples, with the objective of predicting the value taken by f for any given input. In effect, the learning procedure consists in constructing an explicit function that approximates f in some sense. In this article is introduced a Fourier-based machine learning method which could be an alternative or a complement to neural networks for applications in engineering. The basic idea is to extend f into a periodic function so as to use partial sums of the Fourier series as approximations. For this approach to be effective in high dimension, it proved necessary to use several ideas and concepts such as regularization, Sobol sequences and hyperbolic crosses. An attractive feature of the proposed method is that the training stage reduces to a quadratic programming problem. The presented method is first applied to some examples of high-dimensional analytical functions, which allows some comparisons with neural networks to be made. An application to a homogenization problem in nonlinear conduction is discussed in detail. Various examples related to global sensitivity analysis, assessing effective energies of microstructures, and solving boundary value problems are presented.}, - annotation = {WOS:000588550500001}, - author = {Peigney, Michael}, - date = {2021-02-15}, - doi = {10.1002/nme.6565}, - issn = {0029-5981}, - journaltitle = {International Journal for Numerical Methods in Engineering}, - langid = {english}, - location = {{Hoboken}}, - number = {3}, - pages = {866--897}, - publisher = {{Wiley}}, - shortjournal = {Int. J. Numer. Methods Eng.}, - title = {A {{Fourier}}-Based Machine Learning Technique with Application in Engineering}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {122} -} - -@article{pengEfficientLongRangeConvolutions2020, - abstract = {The efficient treatment of long-range interactions for point clouds is a challenging problem in many scientific machine learning applications. To extract global information, one usually needs a large window size, a large number of layers, and/or a large number of channels. This can often significantly increase the computational cost. In this work, we present a novel neural network layer that directly incorporates long-range information for a point cloud. This layer, dubbed the long-range convolutional (LRC)-layer, leverages the convolutional theorem coupled with the non-uniform Fourier transform. In a nutshell, the LRC-layer mollifies the point cloud to an adequately sized regular grid, computes its Fourier transform, multiplies the result by a set of trainable Fourier multipliers, computes the inverse Fourier transform, and finally interpolates the result back to the point cloud. The resulting global all-to-all convolution operation can be performed in nearly-linear time asymptotically with respect to the number of input points. The LRC-layer is a particularly powerful tool when combined with local convolution as together they offer efficient and seamless treatment of both short and long range interactions. We showcase this framework by introducing a neural network architecture that combines LRC-layers with short-range convolutional layers to accurately learn the energy and force associated with a N -body potential. We also exploit the induced two-level decomposition and propose an efficient strategy to train the combined architecture with a reduced number of samples.}, - archiveprefix = {arXiv}, - author = {Peng, Yifan and Lin, Lin and Ying, Lexing and Zepeda-Núñez, Leonardo}, - date = {2020-10-11}, - eprint = {2010.05295}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {cs, math, stat}, - title = {Efficient {{Long}}-{{Range Convolutions}} for {{Point Clouds}}}, - url = {http://arxiv.org/abs/2010.05295}, - urldate = {2021-08-11} -} - -@article{perezSimulationsMeetMachine2018, - abstract = {Classical molecular dynamics (MD) simulations will be able to reach sampling in the second timescale within five years, producing petabytes of simulation data at current force field accuracy. Notwithstanding this, MD will still be in the regime of low-throughput, high-latency predictions with average accuracy. We envisage that machine learning (ML) will be able to solve both the accuracy and time-to-prediction problem by learning predictive models using expensive simulation data. The synergies between classical, quantum simulations and ML methods, such as artificial neural networks, have the potential to drastically reshape the way we make predictions in computational structural biology and drug discovery.}, - archiveprefix = {arXiv}, - author = {Pérez, Adrià and Martínez-Rosell, Gerard and De Fabritiis, Gianni}, - date = {2018-04}, - doi = {10/gdnsnp}, - eprint = {1810.09535}, - eprinttype = {arxiv}, - issn = {0959440X}, - journaltitle = {Current Opinion in Structural Biology}, - langid = {english}, - pages = {139--144}, - shortjournal = {Current Opinion in Structural Biology}, - title = {Simulations Meet {{Machine Learning}} in {{Structural Biology}}}, - url = {http://arxiv.org/abs/1810.09535}, - urldate = {2021-08-11}, - volume = {49} -} - -@article{piaggiEnhancingFormationIonic2021, - abstract = {Ice Ih, the common form of ice in the biosphere, contains proton disorder. Its proton-ordered counterpart, ice XI, is thermodynamically stable below 72 K. However, the formation of ice XI is kinetically hindered, and experimentally it is obtained by doping with KOH. Doping creates ionic defects that promote the migration of protons and the associated change in proton configuration. In this article, we mimic the effect of doping with a bias potential that enhances the formation of ionic defects in molecular dynamics simulations. The recombination of the ions thus formed proceeds through fast migration of the hydroxide along hydrogen bond loops, providing a physical and expedite way to change the proton configuration. A key ingredient of this approach is a machine learning potential trained with density functional theory data and capable of modelling molecular dissociation. We exemplify the usefulness of this idea by studying the order-disorder transition using an appropriate order parameter that distinguishes the proton environments in ice Ih and XI. We calculate the changes in free energy, enthalpy, and entropy associated with the transition. Our estimated entropy agrees with experiment within the error bars of the calculation.}, - annotation = {WOS:000646841900001}, - author = {Piaggi, Pablo M. and Car, Roberto}, - date = {2021-01-22}, - doi = {10.1080/00268976.2021.1916634}, - issn = {0026-8976}, - journaltitle = {Molecular Physics}, - langid = {english}, - location = {{Abingdon}}, - publisher = {{Taylor \& Francis Ltd}}, - shortjournal = {Mol. Phys.}, - title = {Enhancing the Formation of Ionic Defects to Study the Ice {{Ih}}/{{XI}} Transition with Molecular Dynamics Simulations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06} -} - -@article{piaggiPhaseEquilibriumWater2021, - abstract = {Machine learning models are rapidly becoming widely used to simulate complex physicochemical phenomena with ab initio accuracy. Here, we use one such model as well as direct density functional theory (DFT) calculations to investigate the phase equilibrium of water, hexagonal ice (Ih), and cubic ice (Ic), with an eye toward studying ice nucleation. The machine learning model is based on deep neural networks and has been trained on DFT data obtained using the SCAN exchange and correlation functional. We use this model to drive enhanced sampling simulations aimed at calculating a number of complex properties that are out of reach of DFT-driven simulations and then employ an appropriate reweighting procedure to compute the corresponding properties for the SCAN functional. This approach allows us to calculate the melting temperature of both ice polymorphs, the driving force for nucleation, the heat of fusion, the densities at the melting temperature, the relative stability of ices Ih and Ic, and other properties. We find a correct qualitative prediction of all properties of interest. In some cases, quantitative agreement with experiment is better than for state-of-the-art semiempirical potentials for water. Our results also show that SCAN correctly predicts that ice Ih is more stable than ice Ic.}, - annotation = {WOS:000651540200031}, - author = {Piaggi, Pablo M. and Panagiotopoulos, Athanassios Z. and Debenedetti, Pablo G. and Car, Roberto}, - date = {2021-05-11}, - doi = {10.1021/acs.jctc.1c00041}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {5}, - pages = {3065--3077}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {Phase {{Equilibrium}} of {{Water}} with {{Hexagonal}} and {{Cubic Ice Using}} the {{SCAN Functional}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {17} -} - -@article{poltavskyMachineLearningForce2021a, - abstract = {In chemistry and physics, machine learning (ML) methods promise transformative impacts by advancing modeling and improving our understanding of complex molecules and materials. Each ML method comprises a mathematically well-defined procedure, and an increasingly larger number of easy-to-use ML packages for modeling atomistic systems are becoming available. In this Perspective, we discuss the general aspects of ML techniques in the context of creating ML force fields. We describe common features of ML modeling and quantum-mechanical approximations, so-called global and local ML models, and the physical differences behind these two classes of approaches. Finally, we describe the recent developments and emerging directions in the field of ML-driven molecular modeling. This Perspective aims to inspire interdisciplinary collaborations crossing the borders between physical chemistry, chemical physics, computer science, and data science.}, - annotation = {WOS:000677581400014}, - author = {Poltavsky, Igor and Tkatchenko, Alexandre}, - date = {2021-07-22}, - doi = {10.1021/acs.jpclett.1c01204}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {28}, - pages = {6551--6564}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - shorttitle = {Machine {{Learning Force Fields}}}, - title = {Machine {{Learning Force Fields}}: {{Recent Advances}} and {{Remaining Challenges}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {12} -} - -@article{posenitskiyApplicationDeepLearning2021, - abstract = {Deep learning (DL) is applied to simulate non-adiabatic molecular dynamics of phenanthrene, using the time-dependent density functional based tight binding (TD-DFTB) approach for excited states combined with mixed quantum-classical propagation. Reference calculations rely on Tully's fewest-switches surface hopping (FSSH) algorithm coupled to TD-DFTB, which provides electronic relaxation dynamics in fair agreement with various available experimental results. Aiming at describing the coupled electron-nuclei dynamics in large molecular systems, we then examine the combination of DL for excited-state potential energy surfaces (PESs) with a simplified trajectory surface hopping propagation based on the Belyaev-Lebedev (BL) scheme. We start to assess the accuracy of the TD-DFTB approach upon comparison of the optical spectrum with experimental and higher-level theoretical results. Using the recently developed SchNetPack (Schutt et al 2019 J. Chem. Theory Comput. 15 448-55) for DL applications, we train several models and evaluate their performance in predicting excited-state energies and forces. Then, the main focus is given to the analysis of the electronic population of low-lying excited states computed with the aforementioned methods. We determine the relaxation timescales and compare them with experimental data. Our results show that DL demonstrates its ability to describe the excited-state PESs. When coupled to the simplified BL scheme considered in this study, it provides reliable description of the electronic relaxation in phenanthrene as compared with either the experimental data or the higher-level FSSH/TD-DFTB theoretical results. Furthermore, the DL performance allows high-throughput analysis at a negligible cost.}, - annotation = {WOS:000674930500001}, - author = {Posenitskiy, Evgeny and Spiegelman, Fernand and Lemoine, Didier}, - date = {2021-09}, - doi = {10.1088/2632-2153/abfe3f}, - journaltitle = {Machine Learning-Science and Technology}, - langid = {english}, - location = {{Bristol}}, - number = {3}, - pages = {035039}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Mach. Learn.-Sci. Technol.}, - title = {On Application of Deep Learning to Simplified Quantum-Classical Dynamics in Electronically Excited States}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {2} -} - -@article{pulighedduAtomisticSimulationsThermal2020a, - abstract = {We present a method based on sinusoidal approach to equilibrium molecular dynamics (SAEMD) to compute the thermal conductivity of liquids Similar to nonequilibrium molecular dynamics, and unlike equilibrium simulations based on the Green-Kubo formalism, the method only requires the calculation of forces and total energies. The evaluation of heat fluxes and energy densities is not necessary, thus offering the promise of efficiently implementing first principles simulations based on density functional theory or deep molecular dynamics. Our approach is a generalization of SAEMD for solids, where the thermal conductivity is computed in the steady state, instead of a transient regime, thus properly taking into account diffusive terms in the heat equation. We present results for liquid water at ambient conditions and under pressure and discuss simulation requirements to obtain converged values of the thermal conductivity as a function of size and simulation time.}, - annotation = {WOS:000531460800001}, - author = {Puligheddu, Marcello and Galli, Giulia}, - date = {2020-05-11}, - doi = {10.1103/PhysRevMaterials.4.053801}, - issn = {2475-9953}, - journaltitle = {Physical Review Materials}, - langid = {english}, - location = {{College Pk}}, - number = {5}, - pages = {053801}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Mater.}, - title = {Atomistic Simulations of the Thermal Conductivity of Liquids}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {4} -} - -@article{qianComprehensiveAssessmentEmpirical2021, - abstract = {Carbon materials and their unique properties have been extensively studied by molecular dynamics, thanks to the wide range of available carbon bond order potentials (CBOPs). Recently, with the increase in popularity of machine learning (ML), potentials such as Gaussian approximation potential (GAP), trained using ML, can accurately predict results for carbon. However, selecting the right potential is crucial as each performs differently for different carbon allotropes, and these differences can lead to inaccurate results. This work compares the widely used CBOPs and the GAP-20 ML potential with density functional theory results, including lattice constants, cohesive energies, defect formation energies, van der Waals interactions, thermal stabilities, and mechanical properties for different carbon allotropes. We find that GAP-20 can more accurately predict the structure, defect properties, and formation energies for a variety of crystalline phase carbon compared to CBOPs. Importantly, GAP-20 can simulate the thermal stability of C60 and the fracture of carbon nanotubes and graphene accurately, where CBOPs struggle. However, similar to CBOPs, GAP-20 is unable to accurately account for van der Waals interactions. Despite this, we find that GAP-20 outperforms all CBOPs assessed here and is at present the most suitable potential for studying thermal and mechanical properties for pristine and defective carbon.}, - author = {Qian, Cheng and McLean, Ben and Hedman, Daniel and Ding, Feng}, - date = {2021-06}, - doi = {10.1063/5.0052870}, - journaltitle = {APL Materials}, - number = {6}, - publisher = {{American Institute of Physics Inc.}}, - title = {A Comprehensive Assessment of Empirical Potentials for Carbon Materials}, - url = {https://aip.scitation.org/doi/full/10.1063/5.0052870}, - volume = {9} -} - -@article{qiaoOrbNetDeepLearning2020a, - abstract = {We introduce a machine learning method in which energy solutions from the Schrodinger equation are predicted using symmetry adapted atomic orbital features and a graph neural-network architecture. OrbNet is shown to outperform existing methods in terms of learning efficiency and transferability for the prediction of density functional theory results while employing low-cost features that are obtained from semi-empirical electronic structure calculations. For applications to datasets of drug-like molecules, including QM7b-T, QM9, GDB-13-T, DrugBank, and the conformer benchmark dataset of Folmsbee and Hutchison [Int. J. Quantum Chem. (published online) (2020)], OrbNet predicts energies within chemical accuracy of density functional theory at a computational cost that is 1000-fold or more reduced.}, - annotation = {WOS:000577137700001}, - author = {Qiao, Zhuoran and Welborn, Matthew and Anandkumar, Animashree and Manby, Frederick R. and Miller, Thomas F.}, - date = {2020-09-28}, - doi = {10.1063/5.0021955}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {12}, - pages = {124111}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - shorttitle = {{{OrbNet}}}, - title = {{{OrbNet}}: {{Deep}} Learning for Quantum Chemistry Using Symmetry-Adapted Atomic-Orbital Features}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{qiInteractionEnergyPrediction2021a, - abstract = {The interaction energy of two molecules system plays a critical role in analyzing the interacting effect in molecular dynamic simulation. Since the limitation of quantum mechanics calculating resources, the interaction energy based on quantum mechanics can not be merged into molecular dynamic simulation for a long time scale. A deep learning framework, deep tensor neural network, is applied to predict the interaction energy of three organic related systems within the quantum mechanics level of accuracy. The geometric structure and atomic types of molecular conformation, as the data descriptors, are applied as the network inputs to predict the interaction energy in the system. The neural network is trained with the hierarchically generated conformations data set. The complex tensor hidden layers are simplified and trained in the optimization process. The predicted results of different molecular systems indicate that deep tensor neural network is capable to predict the interaction energy with 1 kcal/mol of the mean absolute error in a relatively short time. The prediction highly improves the efficiency of interaction energy calculation. The whole proposed framework provides new insights to introducing deep learning technology into the interaction energy calculation.}, - annotation = {WOS:000629031200012}, - author = {Qi, Yuan and Ren, Hong and Li, Hong and Zhang, Ding-lin and Cui, Hong-qiang and Weng, Jun-ben and Li, Guo-hui and Wang, Gui-yan and Li, Yan}, - date = {2021-02}, - doi = {10.1063/1674-0068/cjcp2009163}, - issn = {1674-0068}, - journaltitle = {Chinese Journal of Chemical Physics}, - langid = {english}, - location = {{Beijing}}, - number = {1}, - pages = {112--124}, - publisher = {{Chinese Physical Soc}}, - shortjournal = {Chin. J. Chem. Phys.}, - title = {Interaction Energy Prediction of Organic Molecules Using Deep Tensor Neural Network}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {34} -} - -@article{ramzanMachineLearningAtomic, - abstract = {We present a new chemically intuitive approach, pairF-Net, to directly predict the atomic forces in a molecule to quantum chemistry accuracy using machine learning techniques. A residual artificial neural network has been designed and trained with features and targets based on pairwise interatomic forces, to determine the Cartesian atomic forces suitable for use in molecular mechanics and dynamics calculations. The scheme implicitly maintains rotational and translational invariance and predicts Cartesian forces as a linear combination of a set of force components in an interatomic basis. We show that the method can predict the reconstructed Cartesian atomic forces for a set of small organic molecules to less than 2 kcal mol-1 Å-1 from the reference force values obtained via density functional theory. The pairF-Net scheme utilises a simple and chemically intuitive route to furnish atomic forces at a quantum mechanical level but at a fraction of the cost, providing a step towards the efficient calculation of accurate thermodynamic properties.}, - author = {Ramzan, I and Kong, L and Bryce, R A and Burton, N A}, - langid = {english}, - pages = {39}, - title = {Machine {{Learning}} of {{Atomic Forces}} from {{Quantum Mechanics}}: A {{Model Based}} on {{Pairwise Interatomic Forces}}}, - url = {https://chemrxiv.org/engage/api-gateway/chemrxiv/assets/orp/resource/item/60c757af567dfe3e20ec66dd/original/machine-learning-of-atomic-forces-from-quantum-mechanics-a-model-based-on-pairwise-interatomic-forces.pdf} -} - -@article{reinhartUnsupervisedLearningAtomic2021a, - abstract = {I present a strategy for unsupervised manifold learning on local atomic environments in molecular simulations based on simple rotation- and permutation-invariant three-body features. These features are highly descriptive, generalize to multiple chemical species, and are human-interpretable. The low-dimensional embeddings of each atomic environment can be used to understand and quantify messy crystal structures such as those near interfaces and defects or well-ordered crystal lattices such as in bulk materials without modification. The same method can also yield collective variables describing collections of particles such as for an entire simulation domain. I demonstrate the method on colloidal crystallization, ice crystals, and binary mesophases to illustrate its broad applicability. In each case, the learned latent space yields insights into the details of the observed microstructures. For ices and mesophases, supervised classifiers are trained based on the learned manifolds and directly compared against a recent neural-network-based approach. Notably, while this method provides comparable classification performance, it can also be deployed on even a handful of observed environments without labels or a priori knowledge. Thus, the current approach provides an incredibly versatile strategy to characterize and classify local atomic environments, and may unlock insights in a wide variety of molecular simulation contexts.}, - annotation = {WOS:000663758800002}, - author = {Reinhart, Wesley F.}, - date = {2021-08}, - doi = {10.1016/j.commatsci.2021.110511}, - issn = {0927-0256}, - journaltitle = {Computational Materials Science}, - langid = {english}, - location = {{Amsterdam}}, - pages = {110511}, - publisher = {{Elsevier}}, - shortjournal = {Comput. Mater. Sci.}, - title = {Unsupervised Learning of Atomic Environments from Simple Features}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {196} -} - -@article{remsingHalogenBondStructure2019, - abstract = {Halogen bonding has emerged as an important noncovalent interaction in a myriad of applications, including drug design, supramolecular assembly, and catalysis. The current understanding of the halogen bond is informed by electronic structure calculations on isolated molecules and/or crystal structures that are not readily transferable to liquids and disordered phases. To address this issue, we present a first-principles simulation-based approach for quantifying halogen bonds in molecular systems rooted in an understanding of nuclei-nuclei and electron-nuclei spatial correlations. We then demonstrate how this approach can be used to quantify the structure and dynamics of halogen bonds in condensed phases, using solid and liquid molecular chlorine as prototypical examples with high concentrations of halogen bonds. We close with a discussion of how the knowledge generated by our first-principles approach may inform the development of classical empirical models, with a consistent representation of halogen bonding.}, - annotation = {WOS:000477786700014}, - author = {Remsing, Richard C. and Klein, Michael L.}, - date = {2019-07-25}, - doi = {10.1021/acs.jpcb.9b04820}, - issn = {1520-6106}, - journaltitle = {Journal of Physical Chemistry B}, - langid = {english}, - location = {{Washington}}, - number = {29}, - pages = {6266--6273}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. B}, - title = {Halogen {{Bond Structure}} and {{Dynamics}} from {{Molecular Simulations}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {123} -} - -@article{renMachineLearningKinetic2021a, - abstract = {Kinetic energy (KE) functional is crucial to speed up density functional theory calculation. However, deriving it accurately through traditional physics reasoning is challenging. We develop a generally applicable KE functional estimator for a one-dimensional (1D) extended system using a machine learning method. Our end-to-end solution combines the dimensionality reduction method with the Gaussian process regression, and simple scaling method to adapt to various 1D lattices. In addition to reaching chemical accuracy in KE calculation, our estimator also performs well on KE functional derivative prediction. Integrating this machine learning KE functional into the current orbital free density functional theory scheme is able to provide us with expected ground state electron density.}, - annotation = {WOS:000658013000001}, - author = {Ren, Hong-Bin and Wang, Lei and Dai, Xi}, - date = {2021-06}, - doi = {10.1088/0256-307X/38/5/050701}, - issn = {0256-307X}, - journaltitle = {Chinese Physics Letters}, - langid = {english}, - location = {{Bristol}}, - number = {5}, - pages = {050701}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Chin. Phys. Lett.}, - title = {Machine {{Learning Kinetic Energy Functional}} for a {{One}}-{{Dimensional Periodic System}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {38} -} - -@article{rodriguezSpatialDensityNeural2020a, - abstract = {Constrained by the fixed mathematical form of most empirical potentials used in classical molecular dynamics (MD) simulations, many properties of materials cannot be captured within experimental accuracy. On the other hand, accurate electronic structure calculations based on quantum theory, most notably density functional theory (DFT), are limited to several hundred atoms within a picosecond, which makes the method inadequate for modeling systems beyond the nanoscale. A combination of speed from classical MD and fidelity from DFT can be achieved through machine learning methods. Herewith, we developed an approach named spatial density neural network force fields (SDNNFFs) by training neural networks to "learn" and predict DFT-level forces. Our model focuses on the usage of a three-dimensional mesh of density functions, which together act as a mapping of the atomic environment and provides a physical representation of the forces acting on the central atom. Several notable advantages arise from the SDNNFF, including (1) the avoidance of the chain rule on the total energy and other variables by direct calculation of the forces from the neural network, (2) the ever large N x t scaling of the training data, where N is the number of atoms in a supercell and t is the number of evaluated structures by first-principles, and (3) the significant reduction in parameters and human effort needed to successfully train a force- and/or property-converged neural network force field. Overall, we focus on modeling DFT-level forces with minimal computational cost and parametrization for rapid prediction of phonon-based properties and future molecular dynamics of large-scale systems. To demonstrate the SDNNFF, we trained several models on diamond structures, including bulk silicon (Si), diamond, silicon carbide (SiC), and boron arsenide (BAs), and predicted their phonon dispersions and lattice thermal conductivities using the direct solution to the phonon Boltzmann transport equation. For phonon properties, we utilized a fitting method for obtaining the second-and third-order force constants, which outperforms the highly force-sensitive finite displacement method when employing neural network force fields. In comparison to DFT lattice thermal conductivity, we obtained high precision results from our SDNNFF within 0.7\% for Si, 6.2\% for diamond, 2.76\% for SiC, and 7.46\% for BAs, with further agreement with experiments. The phonon dispersions from the SDNNFF also matched those from direct DFT and experiments. The developed approach for accurately predicting phonon transport properties of crystalline materials would largely benefit the design of advanced materials with improved performance, such as complex thermoelectric devices and low thermal resistance interfaces for nanoelectronics. Future applications of our SDNNFF model could be extended toward including atomic energy into the algorithm and simulating large-scale heterogeneous systems for quasielectronic representations for various properties.}, - annotation = {WOS:000549756900003}, - author = {Rodriguez, Alejandro and Liu, Yinqiao and Hu, Ming}, - date = {2020-07-17}, - doi = {10.1103/PhysRevB.102.035203}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {3}, - pages = {035203}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Spatial Density Neural Network Force Fields with First-Principles Level Accuracy and Application to Thermal Transport}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {102} -} - -@article{sabbihBiophysicalAnalysisSARSCoV22021a, - abstract = {Recently, SARS-CoV-2 has been identified as the causative factor of viral infection called COVID-19 that belongs to the zoonotic beta coronavirus family known to cause respiratory disorders or viral pneumonia, followed by an extensive attack on organs that express angiotensin-converting enzyme II (ACE2). Human transmission of this virus occurs via respiratory droplets from symptomatic and asymptomatic patients, which are released into the environment after sneezing or coughing. These droplets are capable of staying in the air as aerosols or surfaces and can be transmitted to persons through inhalation or contact with contaminated surfaces. Thus, there is an urgent need for advanced theranostic solutions to control the spread of COVID-19 infection. The development of such fit-for-purpose technologies hinges on a proper understanding of the transmission, incubation, and structural characteristics of the virus in the external environment and within the host. Hence, this article describes the development of an intrinsic model to describe the incubation characteristics of the virus under varying environmental factors. It also discusses on the evaluation of SARS-CoV-2 structural nucleocapsid protein properties via computational approaches to generate high-affinity binding probes for effective diagnosis and targeted treatment applications by specific targeting of viruses. In addition, this article provides useful insights on the transmission behavior of the virus and creates new opportunities for theranostics development.}, - annotation = {WOS:000587567800001}, - author = {Sabbih, Godfred O. and Korsah, Maame A. and Jeevanandam, Jaison and Danquah, Michael K.}, - date = {2021-03}, - doi = {10.1002/btpr.3096}, - issn = {8756-7938}, - journaltitle = {Biotechnology Progress}, - langid = {english}, - location = {{Hoboken}}, - number = {2}, - pages = {e3096}, - publisher = {{Wiley}}, - shortjournal = {Biotechnol. Prog.}, - title = {Biophysical Analysis of {{SARS}}-{{CoV}}-2 Transmission and Theranostic Development via {{N}} Protein Computational Characterization}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {37} -} - -@article{salehActiveLearningPotentialenergy2021, - abstract = {Several pool-based active learning algorithms (AL) were employed to model potential energy surfaces (PESs) with a minimum number of electronic structure calculations. Theoretical and empirical results suggest that superior strategies can be obtained by sampling molecular structures corresponding to large uncertainties in their predictions while at the same time not deviating much from the true distribution of the data. To model PESs in an AL framework we propose to use a regression version of stochastic query by forest, a hybrid method that samples points corresponding to large uncertainties while avoiding collecting too many points from sparse regions of space. The algorithm is implemented with decision trees that come with relatively small computational costs. We empirically show that this algorithm requires around half the data to converge to the same accuracy in comparison to the uncertainty-based query-by-committee algorithm. Moreover, the algorithm is fully automatic and does not require any prior knowledge of the PES. Simulations on a 6D PES of \textbackslash pyrrolew show that \$\textbackslash mathord\{{$<\rbrace$}15\textbackslash,000\$ configurations are enough to build a PES with a generalization error of 16\textasciitilde\textbackslash invcm, whereas the final model with around 50\textbackslash,000 configurations has a generalization error of 11\textasciitilde\textbackslash invcm.\vphantom\}}, - archiveprefix = {arXiv}, - author = {Saleh, Yahya and Sanjay, Vishnu and Iske, Armin and Yachmenev, Andrey and Küpper, Jochen}, - date = {2021-07-15}, - eprint = {2104.00708}, - eprinttype = {arxiv}, - primaryclass = {physics}, - title = {Active Learning of Potential-Energy Surfaces of Weakly-Bound Complexes with Regression-Tree Ensembles}, - url = {http://arxiv.org/abs/2104.00708}, - urldate = {2021-08-11} -} - -@article{sampathClosingGapModeling2020a, - abstract = {Molecular self-assembly is a powerful tool in materials design, wherein noncovalent interactions like electrostatic, hydrophobic, hydrogen bonding, and van der Waals can be exploited to produce supramolecular nanostructures that are functional and highly tunable. Biomolecules are attractive building blocks, as they are biocompatible, biodegradable, and adopt a wide array of higher order structures. Moreover, naturally occurring protein systems display a manifold of structures and interactions that can be replicated in synthetic biomolecules. In this perspective, we highlight advances in multiscale simulation techniques across broad spatiotemporal scales that can aid in characterizing self-assembly of hybrid and hierarchical bionanomaterial systems, with an emphasis on physics-based simulation approaches currently employed to study biomolecules at mineral interfaces. The power of these approaches is highlighted across a few recent areas where molecular simulations have advanced our understanding of self-assembly spanning peptides to protein self-assembly. Looking forward, we discuss how in the near future emerging methods in statistical and machine learning will advance this research field in all areas from expanding the capabilities of physics-based simulation methods to enabling new analyses of high throughput experiments. These advances will pave the way for understanding the molecular recognition patterns in systems that are dictated by self-assembly-biomineralizing peptides, hierarchical peptoids, and large protein assemblies-and will aid in the development of a new synthesis science for achieving precise molecular control in materials design.}, - annotation = {WOS:000580949200001}, - author = {Sampath, Janani and Alamdari, Sarah and Pfaendtner, Jim}, - date = {2020-10-13}, - doi = {10.1021/acs.chemmater.0c01891}, - issn = {0897-4756}, - journaltitle = {Chemistry of Materials}, - langid = {english}, - location = {{Washington}}, - number = {19}, - pages = {8043--8059}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {Chem. Mat.}, - title = {Closing the {{Gap Between Modeling}} and {{Experiments}} in the {{Self}}-{{Assembly}} of {{Biomolecules}} at {{Interfaces}} and in {{Solution}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {32} -} - -@article{saraceniScalableNeuralNetworks2020, - abstract = {Supervised machine learning is emerging as a powerful computational tool to predict the properties of complex quantum systems at a limited computational cost. In this article, we quantify how accurately deep neural networks can learn the properties of disordered quantum systems as a function of the system size. We implement a scalable convolutional network that can address arbitrary system sizes. This network is compared with a recently introduced extensive convolutional architecture [Mills, Chem. Sci. 10, 4129 (2019)2041-652010.1039/C8SC04578J] and with conventional dense networks with all-to-all connectivity. The networks are trained to predict the exact ground-state energies of various disordered systems, namely, a continuous-space single-particle Hamiltonian for cold-atoms in speckle disorder, and different setups of a quantum Ising chain with random couplings, including one with only short-range interactions and one augmented with a long-range term. In all testbeds we consider, the scalable network retains high accuracy as the system size increases. Furthermore, we demonstrate that the network scalability enables a transfer-learning protocol, whereby a pretraining performed on small systems drastically accelerates the learning of large-system properties, allowing reaching high accuracy with small training sets. In fact, with the scalable network one can even extrapolate to sizes larger than those included in the training set, accurately reproducing the results of state-of-the-art quantum Monte Carlo simulations.}, - author = {Saraceni, N. and Cantori, S. and Pilati, S.}, - date = {2020-09}, - doi = {10.1103/physreve.102.033301}, - journaltitle = {Physical Review E}, - number = {3}, - publisher = {{American Physical Society}}, - title = {Scalable Neural Networks for the Efficient Learning of Disordered Quantum Systems}, - url = {https://journals.aps.org/pre/abstract/10.1103/PhysRevE.102.033301}, - volume = {102} -} - -@article{saucedaMolecularForceFields2019, - abstract = {We present the construction of molecular force fields for small molecules (less than 25 atoms) using the recently developed symmetrized gradient-domain machine learning (sGDML) approach [Chmiela et al., Nat. Commun. 9, 3887 (2018); Sci. Adv. 3, e1603015 (2017)]. This approach is able to accurately reconstruct complex high-dimensional potential-energy surfaces from just a few 100s of molecular conformations extracted from ab initio molecular dynamics trajectories. The data efficiency of the sGDML approach implies that atomic forces for these conformations can be computed with high-level wavefunction-based approaches, such as the "gold standard" CCSD(T) method. We demonstrate that the flexible nature of the sGDML model recovers local and non-local electronic interactions (e.g. H-bonding, proton transfer, lone pairs, changes in hybridization states, steric repulsion and \$n\textbackslash to\textbackslash pi\^*\$ interactions) without imposing any restriction on the nature of interatomic potentials. The analysis of sGDML molecular dynamics trajectories yields new qualitative insights into dynamics and spectroscopy of small molecules close to spectroscopic accuracy.}, - archiveprefix = {arXiv}, - author = {Sauceda, Huziel E. and Chmiela, Stefan and Poltavsky, Igor and Müller, Klaus-Robert and Tkatchenko, Alexandre}, - date = {2019-03-21}, - doi = {10/ghqtd7}, - eprint = {1901.06594}, - eprinttype = {arxiv}, - issn = {0021-9606, 1089-7690}, - journaltitle = {The Journal of Chemical Physics}, - langid = {english}, - number = {11}, - pages = {114102}, - shortjournal = {J. Chem. Phys.}, - shorttitle = {Molecular {{Force Fields}} with {{Gradient}}-{{Domain Machine Learning}}}, - title = {Molecular {{Force Fields}} with {{Gradient}}-{{Domain Machine Learning}}: {{Construction}} and {{Application}} to {{Dynamics}} of {{Small Molecules}} with {{Coupled Cluster Forces}}}, - url = {http://arxiv.org/abs/1901.06594}, - urldate = {2021-08-11}, - volume = {150} -} - -@article{schererKernelBasedMachineLearning2020, - abstract = {Current machine learning (ML) models aimed at learning force fields are plagued by their high computational cost at every integration time step. We describe a number of practical and computationally efficient strategies to parametrize traditional force fields for molecular liquids from ML: the particle decomposition ansatz to two- and three-body force fields, the use of kernel-based ML models that incorporate physical symmetries, the incorporation of switching functions close to the cutoff, and the use of covariant meshing to boost the training set size. Results are presented for model molecular liquids: pairwise Lennard-Jones, three-body Stillinger-Weber, and bottom-up coarse-graining of water. Here, covariant meshing proves to be an efficient strategy to learn canonically averaged instantaneous forces. We show that molecular dynamics simulations with tabulated two-and three-body ML potentials are computationally efficient and recover two-and three-body distribution functions. Many-body representations, decomposition, and kernel regression schemes are all implemented in the open-source software package VOTCA.}, - annotation = {WOS:000535226900025}, - author = {Scherer, Christoph and Scheid, Rene and Andrienko, Denis and Bereau, Tristan}, - date = {2020-05-12}, - doi = {10.1021/acs.jctc.9b01256}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {5}, - pages = {3194--3204}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - title = {Kernel-{{Based Machine Learning}} for {{Efficient Simulations}} of {{Molecular Liquids}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {16} -} - -@article{schlederDFTMachineLearning2019, - abstract = {Recent advances in experimental and computational methods are increasing the quantity and complexity of generated data. This massive amount of raw data needs to be stored and interpreted in order to advance the materials science field. Identifying correlations and patterns from large amounts of complex data is being performed by machine learning algorithms for decades. Recently, the materials science community started to invest in these methodologies to extract knowledge and insights from the accumulated data. This review follows a logical sequence starting from density functional theory as the representative instance of electronic structure methods, to the subsequent high-throughput approach, used to generate large amounts of data. Ultimately, data-driven strategies which include data mining, screening, and machine learning techniques, employ the data generated. We show how these approaches to modern computational materials science are being used to uncover complexities and design novel materials with enhanced properties. Finally, we point to the present research problems, challenges, and potential future perspectives of this new exciting field.}, - annotation = {WOS:000560033900003}, - author = {Schleder, Gabriel R. and Padilha, Antonio C. M. and Acosta, Carlos Mera and Costa, Marcio and Fazzio, Adalberto}, - date = {2019-07-01}, - doi = {10.1088/2515-7639/ab084b}, - journaltitle = {Journal of Physics-Materials}, - langid = {english}, - location = {{Bristol}}, - number = {3}, - pages = {032001}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {J. Phys-Mater.}, - shorttitle = {From {{DFT}} to Machine Learning}, - title = {From {{DFT}} to Machine Learning: Recent Approaches to Materials Science-a Review}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {2} -} - -@article{schmidtRecentAdvancesApplications2019a, - abstract = {One of the most exciting tools that have entered the material science toolbox in recent years is machine learning. This collection of statistical methods has already proved to be capable of considerably speeding up both fundamental and applied research. At present, we are witnessing an explosion of works that develop and apply machine learning to solid-state systems. We provide a comprehensive overview and analysis of the most recent research in this topic. As a starting point, we introduce machine learning principles, algorithms, descriptors, and databases in materials science. We continue with the description of different machine learning approaches for the discovery of stable materials and the prediction of their crystal structure. Then we discuss research in numerous quantitative structure-property relationships and various approaches for the replacement of first-principle methods by machine learning. We review how active learning and surrogate-based optimization can be applied to improve the rational design process and related examples of applications. Two major questions are always the interpretability of and the physical understanding gained from machine learning models. We consider therefore the different facets of interpretability and their importance in materials science. Finally, we propose solutions and future research paths for various challenges in computational materials science.}, - annotation = {WOS:000479252200001}, - author = {Schmidt, Jonathan and Marques, Mario R. G. and Botti, Silvana and Marques, Miguel A. L.}, - date = {2019-08-08}, - doi = {10.1038/s41524-019-0221-0}, - journaltitle = {Npj Computational Materials}, - langid = {english}, - location = {{Berlin}}, - pages = {83}, - publisher = {{Nature Research}}, - shortjournal = {npj Comput. Mater.}, - title = {Recent Advances and Applications of Machine Learning in Solid-State Materials Science}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {5} -} - -@article{schranCommitteeNeuralNetwork2020, - abstract = {It is well known in the field of machine learning that committee models improve accuracy, provide generalization error estimates, and enable active learning strategies. In this work, we adapt these concepts to interatomic potentials based on artificial neural networks. Instead of a single model, multiple models that share the same atomic environment descriptors yield an average that outperforms its individual members as well as a measure of the generalization error in the form of the committee disagreement. We not only use this disagreement to identify the most relevant configurations to build up the model's training set in an active learning procedure but also monitor and bias it during simulations to control the generalization error. This facilitates the adaptive development of committee neural network potentials and their training sets while keeping the number of ab initio calculations to a minimum. To illustrate the benefits of this methodology, we apply it to the development of a committee model for water in the condensed phase. Starting from a single reference ab initio simulation, we use active learning to expand into new state points and to describe the quantum nature of the nuclei. The final model, trained on 814 reference calculations, yields excellent results under a range of conditions, from liquid water at ambient and elevated temperatures and pressures to different phases of ice, and the air-water interface-all including nuclear quantum effects. This approach to committee models will enable the systematic development of robust machine learning models for a broad range of systems.}, - annotation = {WOS:000570950400002}, - author = {Schran, Christoph and Brezina, Krystof and Marsalek, Ondrej}, - date = {2020-09-14}, - doi = {10.1063/5.0016004}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {10}, - pages = {104105}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Committee Neural Network Potentials Control Generalization Errors and Enable Active Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{schranTransferabilityMachineLearning2021, - abstract = {A previously published neural network potential for the description of protonated water clusters up to the protonated water tetramer, H+(H2O)(4), at an essentially converged coupled cluster accuracy [C. Schran, J. Behler, and D. Marx, J. Chem. Theory Comput. 16, 88 (2020)] is applied to the protonated water hexamer, H+(H2O)(6)-a system that the neural network has never seen before. Although being in the extrapolation regime, it is shown that the potential not only allows for quantum simulations from ultra-low temperatures similar to 1 K up to 300 K but is also able to describe the new system very accurately compared to explicit coupled cluster calculations. This transferability of the model is rationalized by the similarity of the atomic environments encountered for the larger cluster compared to the environments in the training set of the model. Compared to the interpolation regime, the quality of the model is reduced by roughly one order of magnitude, but most of the difference to the coupled cluster reference comes from global shifts of the potential energy surface, while local energy fluctuations are well recovered. These results suggest that the application of neural network potentials in extrapolation regimes can provide useful results and might be more general than usually thought.}, - annotation = {WOS:000630497100001}, - author = {Schran, Christoph and Brieuc, Fabien and Marx, Dominik}, - date = {2021-02-07}, - doi = {10.1063/5.0035438}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {5}, - pages = {051101}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - shorttitle = {Transferability of Machine Learning Potentials}, - title = {Transferability of Machine Learning Potentials: {{Protonated}} Water Neural Network Potential Applied to the Protonated Water Hexamer}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {154} -} - -@article{schuttSchnetDeepLearning2018, - abstract = {Deep learning has led to a paradigm shift in artificial intelligence, including web, text and image search, speech recognition, as well as bioinformatics, with growing impact in chemical physics. Machine learning in general and deep learning in particular is ideally suited for representing quantum-mechanical interactions, enabling to model nonlinear potential-energy surfaces or enhancing the exploration of chemical compound space. Here we present the deep learning architecture SchNet that is specifically designed to model atomistic systems by making use of continuous-filter convolutional layers. We demonstrate the capabilities of SchNet by accurately predicting a range of properties across chemical space for molecules and materials where our model learns chemically plausible embeddings of atom types across the periodic table. Finally, we employ SchNet to predict potential-energy surfaces and energy-conserving force fields for molecular dynamics simulations of small molecules and perform an exemplary study of the quantum-mechanical properties of C20-fullerene that would have been infeasible with regular ab initio molecular dynamics.}, - author = {Schütt, Kristof T. and Sauceda, Huziel E. and Kindermans, P.-J. and Tkatchenko, Alexandre and Müller, K.-R.}, - date = {2018}, - doi = {10.1063/1.5019779}, - journaltitle = {The Journal of Chemical Physics}, - number = {24}, - pages = {241722}, - publisher = {{AIP Publishing LLC}}, - title = {Schnet–a Deep Learning Architecture for Molecules and Materials}, - url = {https://aip.scitation.org/doi/abs/10.1063/1.5019779}, - volume = {148} -} - -@article{schuttSchNetPackDeepLearning2019, - abstract = {SchNetPack is a toolbox for the development and application of deep neural networks to the prediction of potential energy surfaces and other quantum-chemical properties of molecules and materials. It contains basic building blocks of atomistic neural networks, manages their training and provides simple access to common benchmark datasets. This allows for an easy implementation and evaluation of new models. For now, SchNetPack includes implementations of (weighted) atomcentered symmetry functions and the deep tensor neural network SchNet as well as ready-to-use scripts that allow to train these models on molecule and material datasets. Based upon the PyTorch deep learning framework, SchNetPack allows to efficiently apply the neural networks to large datasets with millions of reference calculations as well as parallelize the model across multiple GPUs. Finally, SchNetPack provides an interface to the Atomic Simulation Environment in order to make trained models easily accessible to researchers that are not yet familiar with neural networks.}, - archiveprefix = {arXiv}, - author = {Schütt, K. T. and Kessel, P. and Gastegger, M. and Nicoli, K. and Tkatchenko, A. and Müller, K.-R.}, - date = {2019-01-08}, - doi = {10/gfrbqm}, - eprint = {1809.01072}, - eprinttype = {arxiv}, - issn = {1549-9618, 1549-9626}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - number = {1}, - pages = {448--455}, - shortjournal = {J. Chem. Theory Comput.}, - shorttitle = {{{SchNetPack}}}, - title = {{{SchNetPack}}: {{A Deep Learning Toolbox For Atomistic Systems}}}, - url = {http://arxiv.org/abs/1809.01072}, - urldate = {2021-08-11}, - volume = {15} -} - -@article{schwalbe-kodaDifferentiableSamplingMolecular2021, - abstract = {Neural network (NN) interatomic potentials provide fast prediction of potential energy surfaces, closely matching the accuracy of the electronic structure methods used to produce the training data. However, NN predictions are only reliable within well-learned training domains, and show volatile behavior when extrapolating. Uncertainty quantification approaches can flag atomic configurations for which prediction confidence is low, but arriving at such uncertain regions requires expensive sampling of the NN phase space, often using atomistic simulations. Here, we exploit automatic differentiation to drive atomistic systems towards high-likelihood, high-uncertainty configurations without the need for molecular dynamics simulations. By performing adversarial attacks on an uncertainty metric, informative geometries that expand the training domain of NNs are sampled. When combined to an active learning loop, this approach bootstraps and improves NN potentials while decreasing the number of calls to the ground truth method. This efficiency is demonstrated on sampling of kinetic barriers and collective variables in molecules, and can be extended to any NN potential architecture and materials system.}, - archiveprefix = {arXiv}, - author = {Schwalbe-Koda, Daniel and Tan, Aik Rui and Gómez-Bombarelli, Rafael}, - date = {2021-03-28}, - eprint = {2101.11588}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 12 pages, 4 figures, supporting information}, - primaryclass = {cond-mat, physics:physics}, - title = {Differentiable Sampling of Molecular Geometries with Uncertainty-Based Adversarial Attacks}, - url = {http://arxiv.org/abs/2101.11588}, - urldate = {2021-08-11} -} - -@article{seoTopologyAutomatedForceField, - abstract = {Force-field development has undergone a revolution in the past decade with the proliferation of quantum chemistry based parameterizations and the introduction of machine learning approximations of the atomistic potential energy surface. Nevertheless, transferable force-fields with broad coverage of organic chemical space remain necessary for applications in materials and chemical discovery where throughput, consistency, and computational cost are paramount. Here we introduce a force-field development framework called Topology Automated Force-Field Interactions (TAFFI) for developing transferable force-fields of varying complexity against an extensible database of quantum chemistry calculations. TAFFI formalizes the concept of atom typing and makes it the basis for generating systematic training data that maintains a one-to-one correspondence with force-field terms. This feature makes TAFFI arbitrarily extensible to new chemistries while maintaining internal consistency and transferability. As a demonstration of TAFFI, we have developed a fixed-charge force-field, TAFFI-gen, from scratch that includes coverage for common organic functional groups that is comparable to established transferable force-fields. The performance of TAFFI-gen was benchmarked against OPLS and GAFF for reproducing several experimental proper- ties of 87 organic liquids. The consistent performance of these force-fields, despite their distinct origins, validates the TAFFI framework while also providing evidence of the representability limitations of fixed-charge force-fields.}, - author = {Seo, Bumjoon and Lin, Zih-Yu and Zhao, Qiyuan and Webb, Michael A and Savoie, M}, - langid = {english}, - pages = {43}, - title = {Topology {{Automated Force}}-{{Field Interactions}} ({{TAFFI}}): {{A Framework}} for {{Developing Transferable Force}}-{{Fields}}}, - url = {https://chemrxiv.org/engage/chemrxiv/article-details/60c7583dbb8c1aad3a3dc9d2} -} - -@article{shangAnharmonicRamanSpectra2021, - abstract = {Raman spectroscopy is an effective tool to analyze the structures of various materials as it provides chemical and compositional information. However, the computation demands for Raman spectra are typically significant because quantum perturbation calculations need to be performed beyond ground state calculations. This work introduces a novel route based on deep neural networks (DNNs) and density-functional perturbation theory to access anharmonic Raman spectra for extended systems. Both the dielectric susceptibility and the potential energy surface are trained using DNNs. The ab initio anharmonic vibrational Raman spectra can be reproduced well with machine learning and DNNs. Silicon and paracetamol crystals are used as showcases to demonstrate the computational efficiency.}, - annotation = {WOS:000630480600005}, - author = {Shang, Honghui and Wang, Haidi}, - date = {2021-03-01}, - doi = {10.1063/5.0040190}, - journaltitle = {Aip Advances}, - langid = {english}, - location = {{Melville}}, - number = {3}, - pages = {035105}, - publisher = {{Amer Inst Physics}}, - shortjournal = {AIP Adv.}, - title = {Anharmonic {{Raman}} Spectra Simulation of Crystals from Deep Neural Networks}, - url = {https://aip.scitation.org/doi/full/10.1063/5.0040190}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{shaoModellingBulkElectrolytes2021, - abstract = {Batteries and supercapacitors are electrochemical energy storage systems which involve multiple time-scales and length-scales. In terms of the electrolyte which serves as the ionic conductor, a molecular-level understanding of the corresponding transport phenomena, electrochemical (thermal) stability and interfacial properties is crucial for optimizing the device performance and achieving safety requirements. To this end, atomistic machine learning is a promising technology for bridging microscopic models and macroscopic phenomena. Here, we provide a timely snapshot of recent advances in this area. This includes technical considerations that are particularly relevant for modelling electrolytes as well as specific examples of both bulk electrolytes and associated interfaces. A perspective on methodological challenges and new applications is also discussed.}, - annotation = {WOS:000604327200001}, - author = {Shao, Yunqi and Knijff, Lisanne and Dietrich, Florian M. and Hermansson, Kersti and Zhang, Chao}, - date = {2021-04}, - doi = {10.1002/batt.202000262}, - journaltitle = {Batteries \& Supercaps}, - langid = {english}, - location = {{Weinheim}}, - number = {4}, - pages = {585--595}, - publisher = {{Wiley-V C H Verlag Gmbh}}, - shortjournal = {Batteries Supercaps}, - title = {Modelling {{Bulk Electrolytes}} and {{Electrolyte Interfaces}} with {{Atomistic Machine Learning}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {4} -} - -@article{shaoPiNNPythonLibrary2020a, - abstract = {Atomic neural networks (ANNs) constitute a class of machine learning methods for predicting potential energy surfaces and physicochemical properties of molecules and materials. Despite many successes, developing interpretable ANN architectures and implementing existing ones efficiently are still challenging. This calls for reliable, general-purpose, and open-source codes. Here, we present a python library named PiNN as a solution toward this goal. In PiNN, we designed a new interpretable and high-performing graph convolutional neural network variant, PiNet, as well as implemented the established Behler-Parrinello neural network. These implementations were tested using datasets of isolated small molecules, crystalline materials, liquid water, and an aqueous alkaline electrolyte. PiNN comes with a visualizer called PiNNBoard to extract chemical insight "learned" by ANNs. It provides analytical stress tensor calculations and interfaces to both the atomic simulation environment and a development version of the Amsterdam Modeling Suite. Moreover, PiNN is highly modularized, which makes it useful not only as a standalone package but also as a chain of tools to develop and to implement novel ANNs. The code is distributed under a permissive BSD license and is freely accessible at https://github.com/Teoroo-CMC/PiNN/with full documentation and tutorials.}, - annotation = {WOS:000526390800014}, - author = {Shao, Yunqi and Hellstrom, Matti and Mitev, Pavlin D. and Knijff, Lisanne and Zhang, Chao}, - date = {2020-03-23}, - doi = {10.1021/acs.jcim.9b00994}, - issn = {1549-9596}, - journaltitle = {Journal of Chemical Information and Modeling}, - langid = {english}, - location = {{Washington}}, - number = {3}, - pages = {1184--1193}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem Inf. Model.}, - shorttitle = {{{PiNN}}}, - title = {{{PiNN}}: {{A Python Library}} for {{Building Atomic Neural Networks}} of {{Molecules}} and {{Materials}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {60} -} - -@article{shapeevElinvarEffectBetaTi2020, - abstract = {A combination of quantum mechanics calculations with machine learning techniques can lead to a paradigm shift in our ability to predict materials properties from first principles. Here we show that on-the-fly training of an interatomic potential described through moment tensors provides the same accuracy as state-of-the-art ab initio molecular dynamics in predicting high-temperature elastic properties of materials with two orders of magnitude less computational effort. Using the technique, we investigate high-temperature bcc phase of titanium and predict very weak, Elinvar, temperature dependence of its elastic moduli, similar to the behavior of the so-called GUM Ti-based alloys (Sato et al 2003 Science 300 464). Given the fact that GUM alloys have complex chemical compositions and operate at room temperature, Elinvar properties of elemental bcc-Ti observed in the wide temperature interval 1100-1700 K is unique.}, - annotation = {WOS:000584988800001}, - author = {Shapeev, Alexander and Podryabinkin, Evgeny and Gubaev, Konstantin and Tasnadi, Ferenc and Abrikosov, Igor A.}, - date = {2020-11}, - doi = {10.1088/1367-2630/abc392}, - issn = {1367-2630}, - journaltitle = {New Journal of Physics}, - langid = {english}, - location = {{Bristol}}, - number = {11}, - pages = {113005}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {New J. Phys.}, - title = {Elinvar Effect in Beta-{{Ti}} Simulated by on-the-Fly Trained Moment Tensor Potential}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {22} -} - -@article{shengPFNNPenaltyfreeNeural2021a, - abstract = {We present PFNN, a penalty-free neural network method, to efficiently solve a class of second-order boundary-value problems on complex geometries. To reduce the smoothness requirement, the original problem is reformulated to a weak form so that the evaluations of high-order derivatives are avoided. Two neural networks, rather than just one, are employed to construct the approximate solution, with one network satisfying the essential boundary conditions and the other handling the rest part of the domain. In this way, an unconstrained optimization problem, instead of a constrained one, is solved without adding any penalty terms. The entanglement of the two networks is eliminated with the help of a length factor function that is scale invariant and can adapt with complex geometries. We prove the convergence of the PFNN method and conduct numerical experiments on a series of linear and nonlinear second-order boundary-value problems to demonstrate that PFNN is superior to several existing approaches in terms of accuracy, flexibility and robustness. (c) 2020 Elsevier Inc. All rights reserved.}, - annotation = {WOS:000612233300005}, - author = {Sheng, Hailong and Yang, Chao}, - date = {2021-03-01}, - doi = {10.1016/j.jcp.2020.110085}, - issn = {0021-9991}, - journaltitle = {Journal of Computational Physics}, - langid = {english}, - location = {{San Diego}}, - pages = {110085}, - publisher = {{Academic Press Inc Elsevier Science}}, - shortjournal = {J. Comput. Phys.}, - shorttitle = {{{PFNN}}}, - title = {{{PFNN}}: {{A}} Penalty-Free Neural Network Method for Solving a Class of Second-Order Boundary-Value Problems on Complex Geometries}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {428} -} - -@article{shenQuantumTrajectoryMeanField2019a, - abstract = {The mixed quantum-classical dynamical approaches have been widely used to study nonadiabatic phenomena in photochemistry and photobiology, in which the time evolutions of the electronic and nuclear subsystems are treated based on quantum and classical mechanics, respectively. The key issue is how to deal with coherence and decoherence during the propagation of the two subsystems, which has been the subject of numerous investigations for a few decades. A brief description on Ehrenfest mean-field and surface-hopping (SH) methods is first provided, and then different algorithms for treatment of quantum decoherence are reviewed in the present paper. More attentions were paid to quantum trajectory meanfield (QTMF) method under the picture of quantum measurements, which is able to overcome the overcoherence problem. Furthermore, the combined QTMF and SH algorithm is proposed in the present work, which takes advantages of the QTMF and SH methods. The potential to extend the applicability of the QTMF method was briefly discussed, such as the generalization to other type of nonadiabatic transitions, the combination with multiscale computational models, and possible improvements on its accuracy and efficiency by using machine-learning techniques.}, - annotation = {WOS:000485829100002}, - author = {Shen, Lin and Tang, Diandong and Xie, Binbin and Fang, Wei-Hai}, - date = {2019-08-29}, - doi = {10.1021/acs.jpca.9b03480}, - issn = {1089-5639}, - journaltitle = {Journal of Physical Chemistry A}, - langid = {english}, - location = {{Washington}}, - number = {34}, - pages = {7337--7350}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. A}, - title = {Quantum {{Trajectory Mean}}-{{Field Method}} for {{Nonadiabatic Dynamics}} in {{Photochemistry}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {123} -} - -@article{Shi2021, - abstract = {Genetic algorithm is widely used to search for the global minimum structure that is important for analyzing the catalyst structure, the mechanism of heterogeneous catalytic reaction, and actual reaction pathway. By performing crossover, mutation and selection, genetic algorithm reflects the process of natural selection where the fittest individuals are selected for reproduction to explore the potential energy surface. As an unbiased optimization algorithm, the optimization process of genetic algorithm does not depend on the input structure and has strong global search capabilities. This review summarizes the recent progress of the design and application of genetic algorithm, as a global structure optimizer, in the catalytic system. Starting with introducing the standard genetic algorithm framework for global structure optimization, this review also includes the advanced framework developed by introducing parallel computing and machine learning technique. Finally, some examples about the reported application of genetic algorithm in catalytic structure optimization are presented, such as the optimization of metallic clusters, supported catalysts, etc. This review might provide a significant insight into the further improvement of genetic algorithm and the wider application in catalytic system.}, - author = {Shi, Xiangcheng and Zhao, Zhijian and Gong, Jinlong}, - date = {2021-01}, - doi = {10.11949/0438-1157.20201037}, - journaltitle = {Huagong Xuebao/CIESC Journal}, - number = {1}, - pages = {27--41}, - publisher = {{Materials China}}, - title = {Application of Genetic Algorithm in the Global Structure Optimization of Catalytic System}, - volume = {72} -} - -@article{shiLearningGradientFields2021, - abstract = {We study a fundamental problem in computational chemistry known as molecular conformation generation, trying to predict stable 3D structures from 2D molecular graphs. Existing machine learning approaches usually first predict distances between atoms and then generate a 3D structure satisfying the distances, where noise in predicted distances may induce extra errors during 3D coordinate generation. Inspired by the traditional force field methods for molecular dynamics simulation, in this paper, we propose a novel approach called ConfGF by directly estimating the gradient fields of the log density of atomic coordinates. The estimated gradient fields allow directly generating stable conformations via Langevin dynamics. However, the problem is very challenging as the gradient fields are rototranslation equivariant. We notice that estimating the gradient fields of atomic coordinates can be translated to estimating the gradient fields of interatomic distances, and hence develop a novel algorithm based on recent score-based generative models to effectively estimate these gradients. Experimental results across multiple tasks show that ConfGF outperforms previous state-ofthe-art baselines by a significant margin. The code is available at https://github.com/ DeepGraphLearning/ConfGF.}, - archiveprefix = {arXiv}, - author = {Shi, Chence and Luo, Shitong and Xu, Minkai and Tang, Jian}, - date = {2021-06-07}, - eprint = {2105.03902}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: ICML 2021, Long talk}, - primaryclass = {physics, q-bio}, - title = {Learning {{Gradient Fields}} for {{Molecular Conformation Generation}}}, - url = {http://arxiv.org/abs/2105.03902}, - urldate = {2021-08-11} -} - -@article{shimamuraComputationalTrainingRequirements2020a, - abstract = {We examined the estimation of thermal conductivity through molecular dynamics simulations for a superionic conductor, alpha-Ag2Se, using the interatomic potential based on an artificial neural network (ANN potential). The training data were created using the existing empirical potential of Ag2Se to help find suitable computational and training requirements for the ANN potential, with the intent to apply them to first-principles calculations. The thermal conductivities calculated using different definitions of heat flux were compared, and the effect of explicit long-range Coulomb interaction on the conductivities was investigated. We clarified that using a rigorous heat flux formula for the ANN potential, even for highly ionic alpha-Ag2Se, the resulting thermal conductivity was reasonably consistent with the reference value without explicitly considering Coulomb interaction. It was found that ANN training including the virial term played an important role in reducing the dependency of thermal conductivity on the initial values of the weight parameters of the ANN.}, - annotation = {WOS:000599772600001}, - author = {Shimamura, Kohei and Takeshita, Yusuke and Fukushima, Shogo and Koura, Akihide and Shimojo, Fuyuki}, - date = {2020-12-21}, - doi = {10.1063/5.0027058}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {23}, - pages = {234301}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Computational and Training Requirements for Interatomic Potential Based on Artificial Neural Network for Estimating Low Thermal Conductivity of Silver Chalcogenides}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{shimamuraEstimatingThermalConductivity2021a, - abstract = {We clarified computational conditions required to construct artificial neural network (ANN) potentials with a Chebyshev descriptor that accurately estimate thermal conductivity (TC). To this end, training data were generated by an empirical potential; α-Ag2Se was selected as the test system and TC was estimated by the homogeneous nonequilibrium molecular dynamics method based on the Green–Kubo formula. The ANN potentials trained with first-principles molecular dynamics data were constructed and the TCs were estimated based on the clarified conditions. Since ANN potentials for multicomponent systems are constructed efficiently with the Chebyshev descriptor, accurate TCs for such systems can be relatively easily estimated.}, - author = {Shimamura, Kohei and Takeshita, Yusuke and Fukushima, Shogo and Koura, Akihide and Shimojo, Fuyuki}, - date = {2021-09-01}, - doi = {10/gj42cx}, - issn = {0009-2614}, - journaltitle = {Chemical Physics Letters}, - langid = {english}, - pages = {138748}, - shortjournal = {Chemical Physics Letters}, - title = {Estimating Thermal Conductivity of α-{{Ag2Se}} Using {{ANN}} Potential with {{Chebyshev}} Descriptor}, - url = {https://www.sciencedirect.com/science/article/pii/S0009261421004310}, - urldate = {2021-08-10}, - volume = {778} -} - -@article{shimamuraGuidelinesCreatingArtificial2019, - abstract = {First-principles molecular dynamics (FPMD) simulations are highly accurate, but due to their high calculation cost, the computational scale is often limited to hundreds of atoms and few picoseconds under specific temperature and pressure conditions. We present here the guidelines for creating artificial neural network empirical interatomic potential (ANN potential) trained with such a limited FPMD data, which can perform long time scale MD simulations at least under the same conditions. The FPMD data for training are prepared on the basis of the convergence of radial distribution function [g(r)]. While training the ANN using total energy and atomic forces of the FPMD data, the error of pressure is also monitored and minimized. To create further robust potential, we add a small amount of FPMD data to reproduce the interaction between two atoms that are close to each other. ANN potentials for alpha-Ag2Se were created as an application example, and it has been confirmed that not only g(r) and mean square displacements but also the specific heat requiring a long time scale simulation matched the FPMD and the experimental values. In addition, the MD simulation using the ANN potential achieved over 10(4) acceleration over the FPMD one. The guidelines proposed here mitigate the creation difficulty of the ANN potential, and a lot of FPMD data sleeping on the hard disk after the research may be put on the front stage again. Published under license by AIP Publishing.}, - annotation = {WOS:000488830300005}, - author = {Shimamura, Kohei and Fukushima, Shogo and Koura, Akihide and Shimojo, Fuyuki and Misawa, Masaaki and Kalia, Rajiv K. and Nakano, Aiichiro and Vashishta, Priya and Matsubara, Takashi and Tanaka, Shigenori}, - date = {2019-09-28}, - doi = {10.1063/1.5116420}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {12}, - pages = {124303}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Guidelines for Creating Artificial Neural Network Empirical Interatomic Potential from First-Principles Molecular Dynamics Data under Specific Conditions and Its Application to Alpha-{{Ag2Se}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {151} -} - -@article{shiWaterDipoleQuadrupole, - abstract = {We report a calculation scheme on water molecular dipole and quadrupole moments in the liquid phase through a Deep Neural Network (DNN) model. Employing the the Maximally Localized Wannier Functions (MLWF) for the valence electrons, we obtain the water moments through a post-process on trajectories from ab-initio molecular dynamics (AIMD) simulations at the density functional theory (DFT) level. In the framework of the deep potential molecular dynamics (DPMD), we develop a scheme to train a DNN with the AIMD moments data. Applying the model, we calculate the contributions from water dipole and quadrupole moments to the electrostatic potential at the center of a cavity of radius 4.1 ˚A as -3.87 V, referenced to the average potential in the bulk-like liquid region. To unravel the ion-independent water effective local potential contribution to the ion hydration free energy, we estimate the 3rd cumulant term as -0.22 V from simulations totally over 6 ns, a time-scale inaccessible for AIMD calculations.}, - author = {Shi, Yu and Doyle, Carrie C and Beck, Thomas L}, - langid = {english}, - pages = {20}, - title = {Water {{Dipole}} and {{Quadrupole Moment Contributions}} to the {{Ion Hydration Free Energy}} by the {{Deep Neural Network}} Trained with {{Ab Initio Molecular Dynamics Data}}} -} - -@article{sinzWaveletScatteringNetworks2020a, - abstract = {The dream of machine learning in materials science is for a model to learn the underlying physics of an atomic system, allowing it to move beyond the interpolation of the training set to the prediction of properties that were not present in the original training data. In addition to advances in machine learning architectures and training techniques, achieving this ambitious goal requires a method to convert a 3D atomic system into a feature representation that preserves rotational and translational symmetries, smoothness under small perturbations, and invariance under re-ordering. The atomic orbital wavelet scattering transform preserves these symmetries by construction and has achieved great success as a featurization method for machine learning energy prediction. Both in small molecules and in the bulk amorphous Li alpha Si system, machine learning models using wavelet scattering coefficients as features have demonstrated a comparable accuracy to density functional theory at a small fraction of the computational cost. In this work, we test the generalizability of our Li alpha Si energy predictor to properties that were not included in the training set, such as elastic constants and migration barriers. We demonstrate that statistical feature selection methods can reduce over-fitting and lead to remarkable accuracy in these extrapolation tasks.}, - annotation = {WOS:000565704800004}, - author = {Sinz, Paul and Swift, Michael W. and Brumwell, Xavier and Liu, Jialin and Kim, Kwang Jin and Qi, Yue and Hirn, Matthew}, - date = {2020-08-28}, - doi = {10.1063/5.0016020}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {8}, - pages = {084109}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Wavelet Scattering Networks for Atomistic Systems with Extrapolation of Material Properties}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{sivaramanExperimentallyDrivenAutomated2021, - abstract = {Understanding the structure and properties of refractory oxides is critical for high temperature applications. In this work, a combined experimental and simulation approach uses an automated closed loop via an active learner, which is initialized by x-ray and neutron diffraction measurements, and sequentially improves a machine-learning model until the experimentally predetermined phase space is covered. A multiphase potential is generated for a canonical example of the archetypal refractory oxide, HfO2, by drawing a minimum number of training configurations from room temperature to the liquid state at ∼2900 °C. The method significantly reduces model development time and human effort.}, - author = {Sivaraman, Ganesh and Gallington, Leighanne and Krishnamoorthy, Anand Narayanan and Stan, Marius and Csányi, Gábor and Vázquez-Mayagoitia, Álvaro and Benmore, Chris J.}, - date = {2021-04-14}, - doi = {10/gkx66f}, - issn = {0031-9007, 1079-7114}, - journaltitle = {Physical Review Letters}, - langid = {english}, - number = {15}, - pages = {156002}, - shortjournal = {Phys. Rev. Lett.}, - title = {Experimentally {{Driven Automated Machine}}-{{Learned Interatomic Potential}} for a {{Refractory Oxide}}}, - url = {https://link.aps.org/doi/10.1103/PhysRevLett.126.156002}, - urldate = {2021-08-11}, - volume = {126} -} - -@article{smithANI1ccxANI1xData2020, - abstract = {Abstract Maximum diversification of data is a central theme in building generalized and accurate machine learning (ML) models. In chemistry, ML has been used to develop models for predicting molecular properties, for example quantum mechanics (QM) calculated potential energy surfaces and atomic charge models. The ANI-1x and ANI-1ccx ML-based general-purpose potentials for organic molecules were developed through active learning; an automated data diversification process. Here, we describe the ANI-1x and ANI-1ccx data sets. To demonstrate data diversity, we visualize it with a dimensionality reduction scheme, and contrast against existing data sets. The ANI-1x data set contains multiple QM properties from 5\,M density functional theory calculations, while the ANI-1ccx data set contains 500\,k data points obtained with an accurate CCSD(T)/CBS extrapolation. Approximately 14 million CPU core-hours were expended to generate this data. Multiple QM calculated properties for the chemical elements C, H, N, and O are provided: energies, atomic forces, multipole moments, atomic charges, etc. We provide this data to the community to aid research and development of ML models for chemistry.}, - author = {Smith, Justin S. and Zubatyuk, Roman and Nebgen, Benjamin and Lubbers, Nicholas and Barros, Kipton and Roitberg, Adrian E. and Isayev, Olexandr and Tretiak, Sergei}, - date = {2020-12}, - doi = {10/gh48xw}, - issn = {2052-4463}, - journaltitle = {Scientific Data}, - langid = {english}, - number = {1}, - pages = {134}, - shortjournal = {Sci Data}, - title = {The {{ANI}}-1ccx and {{ANI}}-1x Data Sets, Coupled-Cluster and Density Functional Theory Properties for Molecules}, - url = {http://www.nature.com/articles/s41597-020-0473-z}, - urldate = {2021-08-11}, - volume = {7} -} - -@article{sommersRamanSpectrumPolarizability2020, - abstract = {We introduce a scheme based on machine learning and deep neural networks to model the environmental dependence of the electronic polarizability in insulating materials. Application to liquid water shows that training the network with a relatively small number of molecular configurations is sufficient to predict the polarizability of arbitrary liquid configurations in close agreement with ab initio density functional theory calculations. In combination with a neural network representation of the interatomic potential energy surface, the scheme allows us to calculate the Raman spectra along 2-nanosecond classical trajectories at different temperatures for H2O and D2O. The vast gains in efficiency provided by the machine learning approach enable longer trajectories and larger system sizes relative to ab initio methods, reducing the statistical error and improving the resolution of the low-frequency Raman spectra. Decomposing the spectra into intramolecular and intermolecular contributions elucidates the mechanisms behind the temperature dependence of the low-frequency and stretch modes.}, - annotation = {WOS:000537251100016}, - author = {Sommers, Grace M. and Andrade, Marcos F. Calegari and Zhang, Linfeng and Wang, Han and Car, Roberto}, - date = {2020-05-21}, - doi = {10.1039/d0cp01893g}, - issn = {1463-9076}, - journaltitle = {Physical Chemistry Chemical Physics}, - langid = {english}, - location = {{Cambridge}}, - number = {19}, - pages = {10592--10602}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Phys. Chem. Chem. Phys.}, - title = {Raman Spectrum and Polarizability of Liquid Water from Deep Neural Networks}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {22} -} - -@article{strickerMachineLearningMetallurgy2020, - abstract = {Interatomic potentials are essential for studying fundamental mechanisms of deformation and failure in metals and alloys because the relevant defects (dislocations, cracks, etc.) are far above the scales accessible to first-principles studies. Existing potentials for non-fcc metals and nearly all alloys are, however, not sufficiently quantitative for many crucial phenomena. Here machine learning in the Behler-Parrinello neural-network framework is used to create a broadly applicable potential for pure hcp magnesium (Mg). Lightweight Mg and its alloys are technologically important while presenting a diverse range of slip systems and crystal surfaces relevant to both plasticity and fracture that present a significant challenge for any potential. The machine learning potential is trained on first-principles density-functional theory (DFT) computable metallurgically relevant properties and is then shown to well predict metallurgically crucial dislocation and crack structures and competing phenomena. Extensive comparisons to an existing very good modified embedded atom method potential are made. These results demonstrate that a single machine learning potential can represent the wide scope of phenomena required for metallurgical studies. The DFT database is openly available for use in any other machine learning method. The method is naturally extendable to alloys, which are necessary for engineering applications but where ductility and fracture are controlled by complex atomic-scale mechanisms that are not well predicted by existing potentials.}, - author = {Stricker, Markus and Yin, Binglun and Mak, Eleanor and Curtin, W. A.}, - date = {2020-10}, - doi = {10.1103/physrevmaterials.4.103602}, - journaltitle = {Physical Review Materials}, - number = {10}, - publisher = {{American Physical Society}}, - title = {Machine Learning for Metallurgy {{II}}. {{A}} Neural-Network Potential for Magnesium}, - volume = {4} -} - -@article{Succi2020, - abstract = {1. Abstract We provide a brief survey of our current developments in the simulation-based design of novel families of mesoscale porous materials using computational kinetic theory. Prospective applications on exascale computers are also briefly discussed and commented on, with reference to two specific examples of soft mesoscale materials: microfluid crystals and bi-continuous jels. 2. Introduction Complex fluid-interface dynamics [1, 2, 3, 4], disordered liquid-liquid emulsions [5, 6], soft-flowing microfluidic crystals [7, 8, 9, 10], all stand as complex states of matter which, besides raising new challenges to modern non-equilibrium thermodynamics, pave the way to many engineering applications, such as combustion and food processing [11, 12], as well as to questions in fundamental biological and physiological processes, like blood flows and protein dynamics [13]. In particular this novel state of soft matter opens up exciting prospects for the design of new materials whose effective building blocks are droplets instead of molecules [14, 7, 15, 16]. The design of new materials has traditionally provided a relentless stimulus to the development of computational schemes spanning the full spectrum of scales, from electrons to atoms and molecules, to supramolecular structures all the way up to the continuum, encompassing over ten decades in space (say from Angstroms to meters) and at least fifteen in time (from femtoseconds to seconds, just to fix ideas). Of course, no single computational model can handle such huge spectrum of scales, each region being treated by dedicated and suitable methods, such as electronic structure simulations, ab-initio molecular dynamics, classical molecular dynamics, stochastic methods, lattice kinetic}, - archiveprefix = {arXiv}, - arxivid = {2005.12764v1}, - author = {Succi, S and Amati, G and Bonaccorso, F and of …, M Lauricella - Journal and 2020, undefined}, - date = {2020}, - eprint = {2005.12764v1}, - eprinttype = {arxiv}, - journaltitle = {Elsevier}, - options = {useprefix=true}, - title = {Toward Exascale Design of Soft Mesoscale Materials}, - url = {https://www.sciencedirect.com/science/article/pii/S1877750320304762} -} - -@article{sugisawaGaussianProcessModel2020a, - abstract = {The goal of the present work is to obtain accurate potential energy surfaces (PESs) for high-dimensional molecular systems with a small number of ab initio calculations in a system-agnostic way. We use probabilistic modeling based on Gaussian processes (GPs). We illustrate that it is possible to build an accurate GP model of a 51-dimensional PES based on 5000 randomly distributed ab initio calculations with a global accuracy of {$<$}0.2 kcal/mol. Our approach uses GP models with composite kernels designed to enhance the Bayesian information content and represents the global PES as a sum of a full-dimensional GP and several GP models for molecular fragments of lower dimensionality. We demonstrate the potency of these algorithms by constructing the global PES for the protonated imidazole dimer, a molecular system with 19 atoms. We illustrate that GP models thus constructed can extrapolate the PES from low energies ({$<$}10 000 cm(-1)), yielding a PES at high energies ({$>$}20 000 cm(-1)). This opens the prospect for new applications of GPs, such as mapping out phase transitions by extrapolation or accelerating Bayesian optimization, for high-dimensional physics and chemistry problems with a restricted number of inputs, i.e., for high-dimensional problems where obtaining training data is very difficult.}, - annotation = {WOS:000573433800001}, - author = {Sugisawa, Hiroki and Ida, Tomonori and Krems, R.}, - date = {2020-09-21}, - doi = {10.1063/5.0023492}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {11}, - pages = {114101}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Gaussian Process Model of 51-Dimensional Potential Energy Surface for Protonated Imidazole Dimer}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{takamotoTeaNetUniversalNeural2019, - abstract = {A universal interatomic potential applicable to arbitrary elements and structures is urgently needed in computational materials science. Graph convolution-based neural network is a promising approach by virtue of its ability to express complex relations. Thus far, it has been thought to represent a completely different approach from physics-based interatomic potentials. In this paper, we show that these two methods can be regarded as different representations of the same tight-binding electronic relaxation framework, where atom-based and overlap integral or "bond"-based Hamiltonian information are propagated in a directional fashion. Based on this unified view, we propose a new model, named the tensor embedded atom network (TeaNet), where the stacked network model is associated with the electronic total energy relaxation calculation. Furthermore, Tersoff-style angular interaction is translated into graph convolution architecture through the incorporation of Euclidean tensor values. Our model can represent and transfer spatial information. TeaNet shows great performance in both the robustness of interatomic potentials and the expressive power of neural networks. We demonstrate that arbitrary chemistry involving the first 18 elements on the periodic table (H to Ar) can be realized by our model, including C-H molecular structures, metals, amorphous SiO\$\{\}\_2\$, and water.}, - archiveprefix = {arXiv}, - author = {Takamoto, So and Izumi, Satoshi and Li, Ju}, - date = {2019-12-02}, - eprint = {1912.01398}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {cond-mat, physics:physics, stat}, - shorttitle = {{{TeaNet}}}, - title = {{{TeaNet}}: Universal Neural Network Interatomic Potential Inspired by Iterative Electronic Relaxations}, - url = {http://arxiv.org/abs/1912.01398}, - urldate = {2021-08-11} -} - -@article{tamInteratomicPotentialSimple2019, - abstract = {Simulations at the atomic scale provide a direct and effective way to understand the mechanical properties of materials. In the regime of classical mechanics, simulations for the thermodynamic properties of metals and alloys can be done by either solving the equations of motion or performing Monte Carlo sampling. The key component for an accurate simulation of such physical systems to produce faithful physical quantities is the use of an appropriate potential or a force field. In this paper, we explore the use of methods from the realm of machine learning to overcome and bypass difficulties encountered when fitting potentials for atomic systems. Particularly, we will show that classical potentials can be represented by a dense neural network with good accuracy.}, - archiveprefix = {arXiv}, - author = {Tam, Ka-Ming and Walker, Nicholas and Kellar, Samuel and Jarrell, Mark}, - date = {2019}, - eprint = {1911.01365}, - eprinttype = {arxiv}, - title = {Interatomic {{Potential}} in a {{Simple Dense Neural Network Representation}}}, - url = {https://arxiv.org/abs/1911.01365} -} - -@article{Tamura2020, - abstract = {We have developed a method that can analyze large random grain boundary (GB) models with the accuracy of density functional theory (DFT) calculations using active learning. It is assumed that the atomic energy is represented by the linear regression of the atomic structural descriptor. The atomic energy is obtained through DFT calculations using a small cell extracted from a huge GB model, called replica DFT atomic energy. The uncertainty reduction (UR) approach in active learning is used to efficiently collect the training data for the atomic energy. In this approach, atomic energy is not required to search for candidate points; therefore, sequential DFT calculations are not required. This approach is suitable for massively parallel computers that can execute a large number of jobs simultaneously. In this study, we demonstrate the prediction of the atomic energy of a Fe random GB model containing one million atoms using the UR approach, and we show that the prediction error decreases more rapidly compared with random sampling. We conclude that the UR approach with replica DFT atomic energy is useful for modeling huge GBs and will be essential for modeling other structural defects.}, - author = {Tamura, Tomoyuki and Karasuyama, Masayuki}, - date = {2020-11}, - doi = {10.1103/physrevmaterials.4.113602}, - journaltitle = {Physical Review Materials}, - number = {11}, - publisher = {{American Physical Society}}, - title = {Prediction of Formation Energies of Large-Scale Disordered Systems via Active-Learning-Based Executions of Ab Initio Local-Energy Calculations: {{A}} Case Study on a {{Fe}} Random Grain Boundary Model with Millions of Atoms}, - volume = {4} -} - -@article{tangChebNetEfficientStable2019, - abstract = {In a recent paper [B. Li, S. Tang and H. Yu, arXiv:1903.05858], it was shown that deep neural networks built with rectified power units (RePU) can give better approximation for sufficient smooth functions than those with rectified linear units, by converting polynomial approximation given in power series into deep neural networks with optimal complexity and no approximation error. However, in practice, power series are not easy to compute. In this paper, we propose a new and more stable way to construct deep RePU neural networks based on Chebyshev polynomial approximations. By using a hierarchical structure of Chebyshev polynomial approximation in frequency domain, we build efficient and stable deep neural network constructions. In theory, ChebNets and the deep RePU nets based on Power series have the same upper error bounds for general function approximations. But numerically, ChebNets are much more stable. Numerical results show that the constructed ChebNets can be further trained and obtain much better results than those obtained by training deep RePU nets constructed basing on power series.}, - archiveprefix = {arXiv}, - author = {Tang, Shanshan and Li, Bo and Yu, Haijun}, - date = {2019-12-20}, - eprint = {1911.05467}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 18 pages, 6 figures, 2 tables}, - primaryclass = {cs, math}, - shorttitle = {{{ChebNet}}}, - title = {{{ChebNet}}: {{Efficient}} and {{Stable Constructions}} of {{Deep Neural Networks}} with {{Rectified Power Units}} Using {{Chebyshev Approximations}}}, - url = {http://arxiv.org/abs/1911.05467}, - urldate = {2021-08-11} -} - -@article{tangDevelopmentInteratomicPotential2020, - abstract = {An interatomic potential for the Al-Tb alloy around the composition of Al(90)Tb(10)is developed using the deep neural network (DNN) learning method. The atomic configurations and the corresponding total potential energies and forces on each atom obtained fromab initiomolecular dynamics (AIMD) simulations are collected to train a DNN model to construct the interatomic potential for the Al-Tb alloy. We show that the obtained DNN model can well reproduce the energies and forces calculated by AIMD simulations. Molecular dynamics (MD) simulations using the DNN interatomic potential also accurately describe the structural properties of the Al(90)Tb(10)liquid, such as partial pair correlation functions (PPCFs) and bond angle distributions, in comparison with the results from AIMD simulations. Furthermore, the developed DNN interatomic potential predicts the formation energies of the crystalline phases of the Al-Tb system with an accuracy comparable toab initiocalculations. The structure factors of the Al(90)Tb(10)metallic liquid and glass obtained by MD simulations using the developed DNN interatomic potential are also in good agreement with the experimental X-ray diffraction data. The development of short-range order (SRO) in the Al(90)Tb(10)liquid and the undercooled liquid is also analyzed and three dominant SROs,i.e., Al-centered distorted icosahedron (DISICO) and Tb-centered '3661' and '15551' clusters, respectively, are identified.}, - annotation = {WOS:000565157900018}, - author = {Tang, L. and Yang, Z. J. and Wen, T. Q. and Ho, K. M. and Kramer, M. J. and Wang, C. Z.}, - date = {2020-09-07}, - doi = {10.1039/d0cp01689f}, - issn = {1463-9076}, - journaltitle = {Physical Chemistry Chemical Physics}, - langid = {english}, - location = {{Cambridge}}, - number = {33}, - pages = {18467--18479}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Phys. Chem. Chem. Phys.}, - title = {Development of Interatomic Potential for {{Al}}-{{Tb}} Alloys Using a Deep Neural Network Learning Method}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {22} -} - -@article{tangShortMediumrangeOrders2021, - abstract = {Molecular dynamics simulations using an interatomic potential developed by artificial neural network deep machine learning are performed to study the local structural order in Al90Tb10 metallic glass. We show that more than 80\% of the Tb-centered clusters in Al90Tb10 glass have short-range order (SRO) with their 17 first coordination shell atoms stacked in a '3661' or '15551' sequence. Medium-range order (MRO) in Bergman-type packing extended out to the second and third coordination shells is also clearly observed. Analysis of the network formed by the '3661' and '15551' clusters show that similar to 82\% of such SRO units share their faces or vertexes, while only similar to 6\% of neighboring SRO pairs are interpenetrating. Such a network topology is consistent with the Bergman-type MRO around the Tb-centers. Moreover, crystal structure searches using genetic algorithm and the neural network interatomic potential reveal several low-energy metastable crystalline structures in the composition range close to Al90Tb10. Some of these crystalline structures have the '3661' SRO while others have the '15551' SRO. While the crystalline structures with the '3661' SRO also exhibit the MRO very similar to that observed in the glass, the ones with the '15551' SRO have very different atomic packing in the second and third shells around the Tb centers from that of the Bergman-type MRO observed in the glassy phase. (c) 2020 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.}, - annotation = {WOS:000603994900021}, - author = {Tang, L. and Yang, Z. J. and Wen, T. Q. and Ho, K. M. and Kramer, M. J. and Wang, C. Z.}, - date = {2021-02-01}, - doi = {10.1016/j.actamat.2020.116513}, - issn = {1359-6454}, - journaltitle = {Acta Materialia}, - langid = {english}, - location = {{Oxford}}, - pages = {116513}, - publisher = {{Pergamon-Elsevier Science Ltd}}, - shortjournal = {Acta Mater.}, - title = {Short- and Medium-Range Orders in {{Al90Tb10}} Glass and Their Relation to the Structures of Competing Crystalline Phases}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {204} -} - -@article{thurstonMachineLearningMolecular2018, - abstract = {Self-assembling oligopeptides present a means to fabricate biocompatible supramolecular aggregates with engineered electronic and optical functionality. We conducted molecular dynamics simulations of self-assembling synthetic oligopeptides with Asp-X cores that mediate hydrophobic stacking and electron delocalisation within the self-assembled nanostructure. The larger PDI cores elevated oligomerisation free energies by a factor of 2-3 relative to NDI and also improved alignment of the oligopeptides within the stack. Training of a quantitative structure-property relationship (QSPR) model revealed key physicochemical determinants of the oligomerisation free energies and produced a predictive model for the oligomerisation thermodynamics. Oligopeptides with moderate dimerisation and trimerisation free energies of produced aggregates with the best in-register parallel stacking, and we used this criterion within our QSPR model to perform high-throughput virtual screening to identify promising candidates for the spontaneous assembly of ordered nanoaggregates. We identified a small number of oligopeptide candidates for direct testing in large scale molecular simulations, and predict a novel chemistry DAVG-PDI-GVAD previously unstudied by experiment or simulation to produce well-aligned nanoaggregates expected to possess good optical and electronic functionality.}, - annotation = {WOS:000433319500007}, - author = {Thurston, Bryce A. and Ferguson, Andrew L.}, - date = {2018}, - doi = {10.1080/08927022.2018.1469754}, - issn = {0892-7022}, - journaltitle = {Molecular Simulation}, - langid = {english}, - location = {{Abingdon}}, - number = {11}, - pages = {930--945}, - publisher = {{Taylor \& Francis Ltd}}, - shortjournal = {Mol. Simul.}, - title = {Machine Learning and Molecular Design of Self-Assembling Pi-Conjugated Oligopeptides}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/6}, - urldate = {2021-08-06}, - volume = {44} -} - -@article{tianRepetitiveLocalSampling, - abstract = {Molecular simulation is a mature and versatile tool set widely utilized in many subjects with more than 30,000 publications each year. However, its methodology development has been struggling with a tradeoff between accuracy/resolution and speed, significant improvement of both beyond present state of the art is necessary to reliably substitute many expensive and laborious experiments in molecular biology, materials science and nanotechnology. Previously, the ubiquitous issue regarding severe wasting of computational resources in all forms of molecular simulations due to repetitive local sampling was raised, and the local free energy landscape approach was proposed to address it. This approach is derived from a simple idea of first learning local distributions, and followed by dynamic assembly of which to infer global joint distribution of a target molecular system. When compared with conventional explicit solvent molecular dynamics simulations, a simple and approximate implementation of this theory in protein structural refinement harvested acceleration of about six orders of magnitude without loss of accuracy. While this initial test revealed tremendous benefits for addressing repetitive local sampling, there are some implicit assumptions need to be articulated. Here, I present a more thorough discussion of repetitive local sampling; potential options for learning local distributions; a more general formulation with potential extension to simulation of near equilibrium molecular systems; the prospect of developing computation driven molecular science; the connection to mainstream residue pair distance distribution based protein structure prediction/refinement; and the fundamental difference of utilizing averaging from conventional molecular simulation framework based on potential of mean force. This more general development is termed the local distribution theory to release the limitation of strict thermodynamic equilibrium in its potential wide application in general soft condensed molecular systems.}, - author = {Tian, Pu}, - langid = {english}, - pages = {32}, - title = {The Repetitive Local Sampling and the Local Distribution Theory}, - url = {https://chemrxiv.org/engage/chemrxiv/article-details/60fd14d4171fc72bafb890fc} -} - -@article{tongCombiningMachineLearning2020, - abstract = {The theoretical structure prediction method via quantum mechanical atomistic simulations such as density functional theory (DFT), based solely on chemical composition, has already become a routine tool to determine the structures of physical and chemical systems, e.g., solids and clusters. However, the application of DFT to more realistic simulations, to a large extent, is impeded because of the unfavorable scaling of the computational cost with respect to the system size. During recent years, the machine learning potential (MLP) method has been rapidly rising as an accurate and efficient tool for atomistic simulations. In this Perspective, we provide an introduction to the basic principles and advantages of the combination of structure prediction and MLP, as well as the challenges and opportunities associated with this promising approach.}, - author = {Tong, Qunchao and Gao, Pengyue and Liu, Hanyu and Xie, Yu and Lv, Jian and Wang, Yanchao and Zhao, Jijun}, - date = {2020-10}, - doi = {10.1021/acs.jpclett.0c02357}, - journaltitle = {Journal of Physical Chemistry Letters}, - number = {20}, - pages = {8710--8720}, - publisher = {{American Chemical Society}}, - title = {Combining Machine Learning Potential and Structure Prediction for Accelerated Materials Design and Discovery}, - url = {https://pubs.acs.org/doi/abs/10.1021/acs.jpclett.0c02357}, - volume = {11} -} - -@article{tongMachineLearningMetadynamics2021, - abstract = {Simulating reconstructive phase transition requires an accurate description of potential energy surface (PES). Density-functional-theory (DFT) based molecular dynamics can achieve the desired accuracy but computationally unfeasible for large systems and/or long simulation times. Here we introduce a new approach that combines the metadynamics simulation and machine learning representation of PES at the accuracy on par with DFT calculation, but with the computational cost several orders of magnitude less, and scaling with system size approximately linear. The high accuracy of the method is demonstrated in the simulation of pressure-induced B4-B1 phase transition in gallium nitride (GaN). The large-scale simulation using a 4096-atom simulation box reveals the phase transition with unprecedented details, including nucleation, growth, and formation of different transition paths under particular stress conditions. With well-trained machine learning potentials, this method can be easily applied to all type of systems for accurate scalable simulations of solid-solid reconstructive phase transition.}, - author = {Tong, Qunchao and Luo, Xiaoshan and Adeleke, Adebayo A. and Gao, Pengyue and Xie, Yu and Liu, Hanyu and Li, Quan and Wang, Yanchao and Lv, Jian and Yao, Yansun and Ma, Yanming}, - date = {2021-02-09}, - doi = {10/gmf5zv}, - issn = {2469-9950, 2469-9969}, - journaltitle = {Physical Review B}, - langid = {english}, - number = {5}, - pages = {054107}, - shortjournal = {Phys. Rev. B}, - title = {Machine Learning Metadynamics Simulation of Reconstructive Phase Transition}, - url = {https://link.aps.org/doi/10.1103/PhysRevB.103.054107}, - urldate = {2021-08-10}, - volume = {103} -} - -@article{townshendGeometricPredictionMoving2020, - abstract = {Many quantities we are interested in predicting are geometric tensors; we refer to this class of problems as geometric prediction. Attempts to perform geometric prediction in real-world scenarios have been limited to approximating them through scalar predictions, leading to losses in data efficiency. In this work, we demonstrate that equivariant networks have the capability to predict real-world geometric tensors without the need for such approximations. We show the applicability of this method to the prediction of force fields and then propose a novel formulation of an important task, biomolecular structure refinement, as a geometric prediction problem, improving state-of-the-art structural candidates. In both settings, we find that our equivariant network is able to generalize to unseen systems, despite having been trained on small sets of examples. This novel and data-efficient ability to predict real-world geometric tensors opens the door to addressing many problems through the lens of geometric prediction, in areas such as 3D vision, robotics, and molecular and structural biology.}, - archiveprefix = {arXiv}, - author = {Townshend, Raphael J. L. and Townshend, Brent and Eismann, Stephan and Dror, Ron O.}, - date = {2020-06-25}, - eprint = {2006.14163}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 10 pages, 8 figures}, - primaryclass = {physics, q-bio, stat}, - shorttitle = {Geometric {{Prediction}}}, - title = {Geometric {{Prediction}}: {{Moving Beyond Scalars}}}, - url = {http://arxiv.org/abs/2006.14163}, - urldate = {2021-08-11} -} - -@article{townshendTransferrableEndtoEndLearning2018, - abstract = {While there has been an explosion in the number of experimentally determined, atomically detailed structures of proteins, how to represent these structures in a machine learning context remains an open research question. In this work we demonstrate that representations learned from raw atomic coordinates can outperform hand-engineered structural features while displaying a much higher degree of transferrability. To do so, we focus on a central problem in biology: predicting how proteins interact with one another—that is, which surfaces of one protein bind to which surfaces of another protein. We present Siamese Atomic Surfacelet Network (SASNet), the first end-to-end learning method for protein interface prediction. Despite using only spatial coordinates and identities of atoms as inputs, SASNet outperforms state-of-the-art methods that rely on hand-engineered, high-level features. These results are particularly striking because we train the method entirely on a significantly biased data set that does not account for the fact that proteins deform when binding to one another. Demonstrating the first successful application of transfer learning to atomic-level data, our network maintains high performance, without retraining, when tested on real cases in which proteins do deform.}, - author = {Townshend, Raphael JL and Bedi, Rishi and Dror, Ron O.}, - date = {2018}, - title = {Transferrable {{End}}-to-{{End Learning}} for {{Protein Interface Prediction}}} -} - -@article{tuoMachineLearningBased2020, - abstract = {A Machine-Learning based Deep Potential (DP) model for Al clusters is developed through training with an extended database including ab initio data of both bulk and several clusters in only 6 CPU/h. This DP model has good performance in accurately predicting the low-lying candidates of Al clusters in a broad size range. Based on our developed DP model, the low-lying structures of 101 different sized Al clusters are extensively searched, among which the lowest-energy candidates of 69 sized clusters are updated. Our calculations demonstrate that machine-learning is indeed powerful in generating potentials to describe the interaction of atoms in complex materials. Published under license by AIP Publishing.}, - annotation = {WOS:000521360300001}, - author = {Tuo, P. and Ye, X. B. and Pan, B. C.}, - date = {2020-03-21}, - doi = {10.1063/5.0001491}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {11}, - pages = {114105}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {A Machine Learning Based Deep Potential for Seeking the Low-Lying Candidates of {{Al}} Clusters}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {152} -} - -@article{unkePhysNetNeuralNetwork2019a, - abstract = {In recent years, machine learning (ML) methods have become increasingly popular in computational chemistry. After being trained on appropriate ab initio reference data, these methods allow for accurately predicting the properties of chemical systems, circumventing the need for explicitly solving the electronic Schrodinger equation. Because of their computational efficiency and scalability to large data sets, deep neural networks (DNNs) are a particularly promising ML algorithm for chemical applications. This work introduces PhysNet, a DNN architecture designed for predicting energies, forces, and dipole moments of chemical systems. PhysNet achieves stateof-the-art performance on the QM9, MD17, and ISO17 benchmarks. Further, two new data sets are generated in order to probe the performance of ML models for describing chemical reactions, long-range interactions, and condensed phase systems. It is shown that explicitly including electrostatics in energy predictions is crucial for a qualitatively correct description of the asymptotic regions of a potential energy surface (PES). PhysNet models trained on a systematically constructed set of small peptide fragments (at most eight heavy atoms) are able to generalize to considerably larger proteins like deca-alanine (Ala(10)): The optimized geometry of helical Ala10 predicted by PhysNet is virtually identical to ab initio results (RMSD = 0.21 A). By running unbiased molecular dynamics (MD) simulations of Ala(10) on the PhysNet-PES in gas phase, it is found that instead of a helical structure, Ala(10) folds into a "wreath-shaped" configuration, which is more stable than the helical form by 0.46 kcal mol(-1) according to the reference ab initio calculations.}, - annotation = {WOS:000471728500021}, - author = {Unke, Oliver T. and Meuwly, Markus}, - date = {2019-06}, - doi = {10.1021/acs.jctc.9b00181}, - issn = {1549-9618}, - journaltitle = {Journal of Chemical Theory and Computation}, - langid = {english}, - location = {{Washington}}, - number = {6}, - pages = {3678--3693}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Chem. Theory Comput.}, - shorttitle = {{{PhysNet}}}, - title = {{{PhysNet}}: {{A Neural Network}} for {{Predicting Energies}}, {{Forces}}, {{Dipole Moments}}, and {{Partial Charges}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {15} -} - -@article{unkeSpookyNetLearningForce2021, - abstract = {Machine-learned force fields (ML-FFs) combine the accuracy of ab initio methods with the efficiency of conventional force fields. However, current ML-FFs typically ignore electronic degrees of freedom, such as the total charge or spin state, and assume chemical locality, which is problematic when molecules have inconsistent electronic states, or when nonlocal effects play a significant role. This work introduces SpookyNet, a deep neural network for constructing ML-FFs with explicit treatment of electronic degrees of freedom and quantum nonlocality. Chemically meaningful inductive biases and analytical corrections built into the network architecture allow it to properly model physical limits. SpookyNet improves upon the current state-of-the-art (or achieves similar performance) on popular quantum chemistry data sets. Notably, it is able to generalize across chemical and conformational space and can leverage the learned chemical insights, e.g. by predicting unknown spin states, thus helping to close a further important remaining gap for today's machine learning models in quantum chemistry.}, - archiveprefix = {arXiv}, - author = {Unke, Oliver T. and Chmiela, Stefan and Gastegger, Michael and Schütt, Kristof T. and Sauceda, Huziel E. and Müller, Klaus-Robert}, - date = {2021-07-20}, - eprint = {2105.00304}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {physics}, - shorttitle = {{{SpookyNet}}}, - title = {{{SpookyNet}}: {{Learning Force Fields}} with {{Electronic Degrees}} of {{Freedom}} and {{Nonlocal Effects}}}, - url = {http://arxiv.org/abs/2105.00304}, - urldate = {2021-08-11} -} - -@article{Vandermause2021, - abstract = {Accurate modeling of chemically reactive systems has traditionally relied on either expensive ab initio approaches or flexible bond-order force fields such as ReaxFF that require considerable time, effort, and expertise to parameterize. Here, we introduce FLARE++, a Bayesian active learning method for training reactive many-body force fields "on the fly" during molecular dynamics (MD) simulations. During the automated training loop, the predictive uncertainties of a sparse Gaussian process (SGP) force field are evaluated at each timestep of an MD simulation to determine whether additional ab initio data are needed. Once trained, the SGP is mapped onto an equivalent and much faster model that is polynomial in the local environment descriptors and whose prediction cost is independent of the training set size. We apply our method to a canonical reactive system in the field of heterogeneous catalysis, hydrogen splitting and recombination on a platinum (111) surface, obtaining a trained model within three days of wall time that is twice as fast as a recent Pt/H ReaxFF force field and considerably more accurate. Our method is fully open source and is expected to reduce the time and effort required to train fast and accurate reactive force fields for complex systems.}, - archiveprefix = {arXiv}, - arxivid = {2106.01949v1}, - author = {Vandermause, J and Xie, Y and Lim, JS and arXiv preprint arXiv …, CJ Owen - and 2021, undefined}, - date = {2021}, - eprint = {2106.01949v1}, - eprinttype = {arxiv}, - isbn = {2106.01949v1}, - journaltitle = {arxiv.org}, - options = {useprefix=true}, - title = {Active Learning of Reactive {{Bayesian}} Force Fields: {{Application}} to Heterogeneous Hydrogen-Platinum Catalysis Dynamics}, - url = {https://arxiv.org/abs/2106.01949} -} - -@article{vandermauseOntheflyActiveLearning2020, - abstract = {Machine learned force fields typically require manual construction of training sets consisting of thousands of first principles calculations, which can result in low training efficiency and unpredictable errors when applied to structures not represented in the training set of the model. This severely limits the practical application of these models in systems with dynamics governed by important rare events, such as chemical reactions and diffusion. We present an adaptive Bayesian inference method for automating the training of interpretable, low-dimensional, and multi-element interatomic force fields using structures drawn on the fly from molecular dynamics simulations. Within an active learning framework, the internal uncertainty of a Gaussian process regression model is used to decide whether to accept the model prediction or to perform a first principles calculation to augment the training set of the model. The method is applied to a range of single- and multi-element systems and shown to achieve a favorable balance of accuracy and computational efficiency, while requiring a minimal amount of ab initio training data. We provide a fully open-source implementation of our method, as well as a procedure to map trained models to computationally efficient tabulated force fields.}, - annotation = {WOS:000520616900001}, - author = {Vandermause, Jonathan and Torrisi, Steven B. and Batzner, Simon and Xie, Yu and Sun, Lixin and Kolpak, Alexie M. and Kozinsky, Boris}, - date = {2020-03-18}, - doi = {10.1038/s41524-020-0283-z}, - journaltitle = {Npj Computational Materials}, - langid = {english}, - location = {{London}}, - number = {1}, - pages = {20}, - publisher = {{Nature Publishing Group}}, - shortjournal = {npj Comput. Mater.}, - title = {On-the-Fly Active Learning of Interpretable {{Bayesian}} Force Fields for Atomistic Rare Events}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {6} -} - -@article{vandermauseOntheflyBayesianActive2019, - abstract = {Machine learning based interatomic potentials currently require manual construction of training sets consisting of thousands of first principles calculations and are often restricted to single-component and nonreactive systems. This severely limits the practical application of these models due to both low training efficiency and limited accuracy in treating important rare events such as reactions and diffusion. We present an adaptive Bayesian inference method for automating and accelerating the on-the-fly construction of accurate interatomic force fields using structures drawn from molecular dynamics simulations. Within an online active learning algorithm, the internal uncertainty of a Gaussian process regression model is used to decide whether to accept the model prediction or to perform a first principles calculation to augment the training set of the model. The method is applied to a range of single-and multi-component systems and shown to achieve state-of-the-art accuracy with minimal ab initio data.}, - archiveprefix = {arXiv}, - arxivid = {1904.02042v2}, - author = {Vandermause, J and Torrisi, SB and Batzner, S}, - date = {2019}, - eprint = {1904.02042v2}, - eprinttype = {arxiv}, - journaltitle = {projects.iq.harvard.edu}, - title = {On-the-Fly {{Bayesian}} Active Learning of Interpretable Force-Fields for Atomistic Rare Events}, - url = {https://projects.iq.harvard.edu/files/bkoz/files/2_vandermause_et_al._-_2019_-_on-the-fly_bayesian_active_learning_of_interpretable_force-fields_for_atomistic_rare_events.pdf} -} - -@article{vassilev-galindoChallengesMachineLearning2021, - abstract = {Dynamics of flexible molecules are often determined by an interplay between local chemical bond fluctuations and conformational changes driven by long-range electrostatics and van der Waals interactions. This interplay between interactions yields complex potential-energy surfaces (PESs) with multiple minima and transition paths between them. In this work, we assess the performance of the state-of-the-art Machine Learning (ML) models, namely, sGDML, SchNet, Gaussian Approximation Potentials/Smooth Overlap of Atomic Positions (GAPs/SOAPs), and Behler-Parrinello neural networks, for reproducing such PESs, while using limited amounts of reference data. As a benchmark, we use the cis to trans thermal relaxation in an azobenzene molecule, where at least three different transition mechanisms should be considered. Although GAP/SOAP, SchNet, and sGDML models can globally achieve a chemical accuracy of 1 kcal mol(-1) with fewer than 1000 training points, predictions greatly depend on the ML method used and on the local region of the PES being sampled. Within a given ML method, large differences can be found between predictions of close-to-equilibrium and transition regions, as well as for different transition mechanisms. We identify key challenges that the ML models face mainly due to the intrinsic limitations of commonly used atom-based descriptors. All in all, our results suggest switching from learning the entire PES within a single model to using multiple local models with optimized descriptors, training sets, and architectures for different parts of the complex PES.}, - annotation = {WOS:000630524000019}, - author = {Vassilev-Galindo, Valentin and Fonseca, Gregory and Poltavsky, Igor and Tkatchenko, Alexandre}, - date = {2021-03-07}, - doi = {10.1063/5.0038516}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {9}, - pages = {094119}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Challenges for Machine Learning Force Fields in Reproducing Potential Energy Surfaces of Flexible Molecules}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {154} -} - -@article{venturiBayesianMachineLearning2020, - abstract = {This work introduces a novel methodology for the quantification of uncertainties associated with potential energy surfaces (PESs) computed from first-principles quantum mechanical calculations. The methodology relies on Bayesian inference and machine learning techniques to construct a stochastic PES and to express the inadequacies associated with the ab initio data points and their fit. By combining high fidelity calculations and reduced-order modeling, the resulting stochastic surface is efficiently forward propagated via quasi-classical trajectory and master equation calculations. In this way, the PES contribution to the uncertainty on predefined quantities of interest (QoIs) is explicitly determined. This study is done at both microscopic (e.g., rovibrational-specific rate coefficients) and macroscopic (e.g., thermal and chemical relaxation properties) levels. A correlation analysis is finally applied to identify the PES regions that require further refinement, based on their effects on the QoI reliability. The methodology is applied to the study of singlet (11A′) and quintet (25A′) PESs describing the interaction between O2 molecules and O atoms in their ground electronic state. The investigation of the singlet surface reveals a negligible uncertainty on the kinetic properties and relaxation times, which are found to be in excellent agreement with the ones previously published in the literature. On the other hand, the methodology demonstrated significant uncertainty on the quintet surface, due to inaccuracies in the description of the exchange barrier and the repulsive wall. When forward propagated, this uncertainty is responsible for the variability of 1 order of magnitude in the vibrational relaxation time and of factor four in the exchange reaction rate coefficient, both at 2500 K.}, - author = {Venturi, S. and Jaffe, R. L. and Panesi, M.}, - date = {2020-06}, - doi = {10.1021/acs.jpca.0c02395}, - journaltitle = {Journal of Physical Chemistry A}, - number = {25}, - pages = {5129--5146}, - publisher = {{American Chemical Society}}, - title = {Bayesian Machine Learning Approach to the Quantification of Uncertainties on Ab Initio Potential Energy Surfaces}, - volume = {124} -} - -@article{vergadouMolecularModelingInvestigations2019a, - abstract = {With a wide range of applications, from energy and environmental engineering, such as in gas separations and water purification, to biomedical engineering and packaging, glassy polymeric materials remain in the core of novel membrane and state-of the art barrier technologies. This review focuses on molecular simulation methodologies implemented for the study of sorption and diffusion of small molecules in dense glassy polymeric systems. Basic concepts are introduced and systematic methods for the generation of realistic polymer configurations are briefly presented. Challenges related to the long length and time scale phenomena that govern the permeation process in the glassy polymer matrix are described and molecular simulation approaches developed to address the multiscale problem at hand are discussed.}, - annotation = {WOS:000482952800009}, - author = {Vergadou, Niki and Theodorou, Doros N.}, - date = {2019-08}, - doi = {10.3390/membranes9080098}, - journaltitle = {Membranes}, - langid = {english}, - location = {{Basel}}, - number = {8}, - pages = {98}, - publisher = {{Mdpi}}, - shortjournal = {Membranes}, - title = {Molecular {{Modeling Investigations}} of {{Sorption}} and {{Diffusion}} of {{Small Molecules}} in {{Glassy Polymers}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {9} -} - -@article{vinsonFasterExactExchange2020a, - abstract = {Density-functional theory simplifies many-electron calculations by approximating the exchange and correlation interactions with a one-electron operator that is a functional of the density. Hybrid functionals incorporate some amount of exact exchange, improving agreement with measured electronic and structural properties. However, calculations with hybrid functionals require substantial computational resources, limiting their use. By calculating the exchange interaction of periodic systems with single-precision arithmetic, the computation time is cut nearly in half with a negligible loss in accuracy. This improvement makes exact exchange calculations quicker and more feasible, especially for high-throughput calculations. Example hybrid density-functional theory calculations of band energies, forces, and x-ray absorption spectra show that this single-precision implementation maintains accuracy with significantly reduced runtime and memory requirements.}, - annotation = {WOS:000595858100002}, - author = {Vinson, John}, - date = {2020-11-28}, - doi = {10.1063/5.0030493}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {20}, - pages = {204106}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Faster Exact Exchange in Periodic Systems Using Single-Precision Arithmetic}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{wangCombiningFragmentationApproach2020a, - abstract = {Accurate and efficient all-atom quantum mechanical (QM) calculations for biomolecules still present a challenge to computational physicists and chemists. In this study, an extensible generalized molecular fractionation with a conjugate caps method combined with neural networks (NN-GMFCC) is developed for efficient QM calculation of protein energy. In the NN-GMFCC scheme, the total energy of a given protein is calculated by taking a proper combination of the high-precision neural network potential energies of all capped residues and overlapping conjugate caps. In addition, the two-body interaction energies of residue pairs are calculated by molecular mechanics (MM). With reference to the GMFCC/MM calculation at the.B97XD/6-31G* level, the overall mean unsigned errors of the energy deviations and atomic force root-mean-squared errors calculated by NN-GMFCC are only 2.01 kcal/mol and 0.68 kcal/mol/A, respectively, for 14 proteins (containing up to 13,728 atoms). Meanwhile, the NN-GMFCC approach is about 4 orders of magnitude faster than the GMFCC/MM method. The NN-GMFCC method could be systematically improved by inclusion of two-body QM interaction and multibody electronic polarization effect. Moreover, the NN-GMFCC approach can also be applied to other macromolecular systems such as DNA/RNA, and it is capable of providing a powerful and efficient approach for exploration of structures and functions of proteins with QM accuracy.}, - annotation = {WOS:000526368900006}, - author = {Wang, Zhilong and Han, Yanqiang and Li, Jinjin and He, Xiao}, - date = {2020-04-16}, - doi = {10.1021/acs.jpcb.0c01370}, - issn = {1520-6106}, - journaltitle = {Journal of Physical Chemistry B}, - langid = {english}, - location = {{Washington}}, - number = {15}, - pages = {3027--3035}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. B}, - title = {Combining the {{Fragmentation Approach}} and {{Neural Network Potential Energy Surfaces}} of {{Fragments}} for {{Accurate Calculation}} of {{Protein Energy}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {124} -} - -@article{wangComplexReactionNetwork2021, - abstract = {An approach based on ab initio statistical mechanics is demonstrated for autoconstructing complex reaction networks. Ab initio molecular dynamics combined with Markov state models are employed to study relevant transitions and corresponding thermodynamic and kinetic properties of a reaction. To explore the capability and flexibility of this approach, we present a study of oxygen activation on Ag4 as a model reaction. Specifically, with the same sampled trajectories, it is possible to study the structural effects and the reaction rate of the cited reaction. The results show that this approach is suitable for automatized construction of reaction networks, especially for non-well-studied reactions, which can benefit from this ab initio molecular dynamics based approach to construct comprehensive reaction networks with Markov state models without prior knowledge about the potential energy landscape.}, - author = {Wang, Weiqi and Liu, Xiangyue and Pérez-Ríos, Jesús}, - date = {2021-07-01}, - doi = {10/gmfw5m}, - issn = {1089-5639, 1520-5215}, - journaltitle = {The Journal of Physical Chemistry A}, - langid = {english}, - number = {25}, - pages = {5670--5680}, - shortjournal = {J. Phys. Chem. A}, - shorttitle = {Complex {{Reaction Network Thermodynamic}} and {{Kinetic Autoconstruction Based}} on {{{\emph{Ab Initio}}}} {{Statistical Mechanics}}}, - title = {Complex {{Reaction Network Thermodynamic}} and {{Kinetic Autoconstruction Based}} on {{{\emph{Ab Initio}}}} {{Statistical Mechanics}}: {{A Case Study}} of {{O}} {\textsubscript{2}} {{Activation}} on {{Ag}} {\textsubscript{4}} {{Clusters}}}, - url = {https://pubs.acs.org/doi/10.1021/acs.jpca.1c03454}, - urldate = {2021-08-11}, - volume = {125} -} - -@article{wangCrystalStructurePrediction2020, - abstract = {Predicting crystal structure has been a challenging problem in physics and materials science for a long time. A reliable energy calculation engine combined with an efficient global search algorithm, such as particle swarm optimization algorithm or genetic algorithm, is needed to conduct crystal structure prediction. In recent years, machine learning-based interatomic potential energy surface models have been proposed, potentially allowing us to perform crystal structure prediction for systems with the accuracy of density functional theory (DFT) and the speed of empirical force fields. In this paper, we employ a previously developed Deep Potential model to predict the intermetallic compound of the aluminum-magnesium system, and find six meta-stable phases with negative or nearly zero formation energy. In particular, Mg12Al8 shows excellent ductility and Mg5Al27 has a high Young's modulus. Based on our benchmark results, we propose a relatively robust structure screening criterion that selects potentially stable structures from the Deep Potential-based convex hull and performs DFT refinement. By using this criterion, the computational cost needed to construct the convex hull with ab initio accuracy can be dramatically reduced.}, - annotation = {WOS:000596834900001}, - author = {Wang, Haidi and Zhang, Yuzhi and Zhang, Linfeng and Wang, Han}, - date = {2020-11-26}, - doi = {10.3389/fchem.2020.589795}, - issn = {2296-2646}, - journaltitle = {Frontiers in Chemistry}, - langid = {english}, - location = {{Lausanne}}, - pages = {589795}, - publisher = {{Frontiers Media Sa}}, - shortjournal = {Front. Chem.}, - title = {Crystal {{Structure Prediction}} of {{Binary Alloys}} via {{Deep Potential}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {8} -} - -@article{wangDeepLearningInteratomic2019, - abstract = {We propose a hybrid scheme that smoothly interpolates the Ziegler-Biersack-Littmark (ZBL) screened nuclear repulsion potential with a deep learning potential energy model. The resulting deep potential-ZBL model can not only provide overall good performance on the predictions of near-equilibrium material properties but also capture the right physics when atoms are extremely close to each other, an event that frequently happens in computational simulations of irradiation damage events. We applied this scheme to the simulation of the irradiation damage processes in the face-centered-cubic aluminum system and found better descriptions in terms of the defect formation energy, evolution of collision cascades, displacement threshold energy, and residual point defects than the widely adopted ZBL modified embedded atom method potentials and their variants. Our work provides a reliable and feasible scheme to accurately simulate the irradiation damage processes and opens up extra opportunities to solve the predicament of lacking accurate potentials for enormous recently discovered materials in the irradiation effect field.}, - annotation = {WOS:000472599100024}, - author = {Wang, Hao and Guo, Xun and Zhang, Linfeng and Wang, Han and Xue, Jianming}, - date = {2019-06-17}, - doi = {10.1063/1.5098061}, - issn = {0003-6951}, - journaltitle = {Applied Physics Letters}, - langid = {english}, - location = {{Melville}}, - number = {24}, - pages = {244101}, - publisher = {{Amer Inst Physics}}, - shortjournal = {Appl. Phys. Lett.}, - title = {Deep Learning Inter-Atomic Potential Model for Accurate Irradiation Damage Simulations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {114} -} - -@article{wangDeeplearningInteratomicPotential2020, - abstract = {Potentials that could accurately describe the irradiation damage processes are highly desired to figure out the atomic-level response of various newly-discovered materials under irradiation environments. In this work, we introduce a deep-learning interatomic potential for monolayer MoS2 by combining all-electron calculations, an active-learning sampling method and a hybrid deep-learning model. This potential could not only give an overall good performance on the predictions of near-equilibrium material properties including lattice constants, elastic coefficients, energy stress curves, phonon spectra, defect formation energy and displacement threshold, but also reproduce the ab initial irradiation damage processes with high quality. Further irradiation simulations indicate that one single highenergy ion could generate a large nanopore with a diameter of more than 2 nm, or a series of multiple nanopores, which is qualitatively verified by the subsequent 500 keV Au+ ion irradiation experiments. This work provides a promising and feasible approach to simulate irradiation effects in enormous newly-discovered materials with unprecedented accuracy.}, - archiveprefix = {arXiv}, - author = {Wang, Hao and Guo, Xun and Xue, Jianming}, - date = {2020-10-19}, - eprint = {2010.09547}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {cond-mat, physics:physics}, - title = {Deep-Learning Interatomic Potential for Irradiation Damage Simulations in {{MoS2}} with Ab Initial Accuracy}, - url = {http://arxiv.org/abs/2010.09547}, - urldate = {2021-08-11} -} - -@article{wangDeePMDkitDeepLearning2018, - abstract = {Recent developments in many-body potential energy representation via deep learning have brought new hopes to addressing the accuracy-versus-efficiency dilemma in molecular simulations. Here we describe DeePMD-kit, a package written in Python/C++ that has been designed to minimize the effort required to build deep learning based representation of potential energy and force field and to perform molecular dynamics. Potential applications of DeePMD-kit span from finite molecules to extended systems and from metallic systems to chemically bonded systems. DeePMD-kit is interfaced with TensorFlow, one of the most popular deep learning frameworks, making the training process highly automatic and efficient. On the other end, DeePMD-kit is interfaced with high-performance classical molecular dynamics and quantum (path-integral) molecular dynamics packages, i.e., LAMMPS and the i-PI, respectively. Thus, upon training, the potential energy and force field models can be used to perform efficient molecular simulations for different purposes. As an example of the many potential applications of the package, we use DeePMD-kit to learn the interatomic potential energy and forces of a water model using data obtained from density functional theory. We demonstrate that the resulted molecular dynamics model reproduces accurately the structural information contained in the original model. Program summary Program Title: DeePMD-kit Program Files doi: http://dx.doi.org/10.17632/hvfh9yvncf.1 Licensing provisions: LGPL Programming language: Python/C++ Nature of problem: Modeling the many-body atomic interactions by deep neural network models. Running molecular dynamics simulations with the models. Solution method: The Deep Potential for Molecular Dynamics (DeePMD) method is implemented based on the deep learning framework TensorFlow. Supports for using a DeePMD model in LAMMPS and i-PI, for classical and quantum (path integral) molecular dynamics are provided. Additional comments including Restrictions and Unusual features: The code defines a data protocol such that the energy, force, and virial calculated by different third-party molecular simulation packages can be easily processed and used as model training data. (C) 2018 Elsevier B.V. All rights reserved.}, - annotation = {WOS:000434000900019}, - author = {Wang, Han and Zhang, Linfeng and Han, Jiequn and E, Weinan}, - date = {2018-07}, - doi = {10.1016/j.cpc.2018.03.016}, - issn = {0010-4655}, - journaltitle = {Computer Physics Communications}, - langid = {english}, - location = {{Amsterdam}}, - pages = {178--184}, - publisher = {{Elsevier Science Bv}}, - shortjournal = {Comput. Phys. Commun.}, - shorttitle = {{{DeePMD}}-Kit}, - title = {{{DeePMD}}-Kit: {{A}} Deep Learning Package for Many-Body Potential Energy Representation and Molecular Dynamics}, - url = {https://www.sciencedirect.com/science/article/abs/pii/S0010465518300882}, - urldate = {2021-08-05}, - volume = {228} -} - -@article{wangDifferentiableMolecularSimulations2020, - abstract = {Molecular dynamics simulations use statistical mechanics at the atomistic scale to enable both the elucidation of fundamental mechanisms and the engineering of matter for desired tasks. The behavior of molecular systems at the microscale is typically simulated with differential equations parameterized by a Hamiltonian, or energy function. The Hamiltonian describes the state of the system and its interactions with the environment. In order to derive predictive microscopic models, one wishes to infer a molecular Hamiltonian that agrees with observed macroscopic quantities. From the perspective of engineering, one wishes to control the Hamiltonian to achieve desired simulation outcomes and structures, as in self-assembly and optical control, to then realize systems with the desired Hamiltonian in the lab. In both cases, the goal is to modify the Hamiltonian such that emergent properties of the simulated system match a given target. We demonstrate how this can be achieved using differentiable simulations where bulk target observables and simulation outcomes can be analytically differentiated with respect to Hamiltonians, opening up new routes for parameterizing Hamiltonians to infer macroscopic models and develop control protocols.}, - archiveprefix = {arXiv}, - author = {Wang, Wujie and Axelrod, Simon and Gómez-Bombarelli, Rafael}, - date = {2020-12-23}, - eprint = {2003.00868}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 14 pages, 6 figures}, - primaryclass = {physics, stat}, - title = {Differentiable {{Molecular Simulations}} for {{Control}} and {{Learning}}}, - url = {http://arxiv.org/abs/2003.00868}, - urldate = {2021-08-11} -} - -@article{wangElectronicallyDriven1D2021a, - abstract = {Atomic diffusion is a spontaneous process and significantly influences properties of materials, such as fracture toughness, creep-fatigue properties, thermal conductivity, thermoelectric properties, etc. Here, using extensive molecular dynamics simulations based on both ab initio and machine-learning potentials, we demonstrate that an atomic one dimensional cooperative diffusion exists in the simple cubic high-pressure finite-temperature phase of calcium in the premelting regime, where some atoms diffuse cooperatively as chains or even rings, while others remain in the solid state. This intermediate regime is triggered by anharmonicity of the system at high temperature and is stabilized by the competition between the internal energy minimization and the entropy maximization, and has close connections with the unique electronic structures of simple cubic Ca as an electride with a pseudogap. This cooperative diffusion regime explains the abnormal enhancement of the melting line of Ca under high pressure and suggests that the cooperative chain melting is a much more common high-temperature feature among metals under extreme conditions than hitherto thought. The microscopic electronic investigations of these systems combining ab initio and machine-learning data point out the direction for further understanding of other metallic systems such as the glass transition, liquid metals, etc.}, - annotation = {WOS:000606733200001}, - author = {Wang, Yong and Wang, Junjie and Hermann, Andreas and Liu, Cong and Gao, Hao and Tosatti, Erio and Wang, Hui-Tian and Xing, Dingyu and Sun, Jian}, - date = {2021-01-11}, - doi = {10.1103/PhysRevX.11.011006}, - issn = {2160-3308}, - journaltitle = {Physical Review X}, - langid = {english}, - location = {{College Pk}}, - number = {1}, - pages = {011006}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. X}, - title = {Electronically {{Driven 1D Cooperative Diffusion}} in a {{Simple Cubic Crystal}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{wangEnsembleLearningCoarsegrained2020a, - abstract = {Gradient-domain machine learning (GDML) is an accurate and efficient approach to learn a molecular potential and associated force field based on the kernel ridge regression algorithm. Here, we demonstrate its application to learn an effective coarse-grained (CG) model from all-atom simulation data in a sample efficient manner. The CG force field is learned by following the thermodynamic consistency principle, here by minimizing the error between the predicted CG force and the all-atom mean force in the CG coordinates. Solving this problem by GDML directly is impossible because coarse-graining requires averaging over many training data points, resulting in impractical memory requirements for storing the kernel matrices. In this work, we propose a data-efficient and memory-saving alternative. Using ensemble learning and stratified sampling, we propose a 2-layer training scheme that enables GDML to learn an effective CG model. We illustrate our method on a simple biomolecular system, alanine dipeptide, by reconstructing the free energy landscape of a CG variant of this molecule. Our novel GDML training scheme yields a smaller free energy error than neural networks when the training set is small, and a comparably high accuracy when the training set is sufficiently large.}, - annotation = {WOS:000536701800001}, - author = {Wang, Jiang and Chmiela, Stefan and Mueller, Klaus-Robert and Noe, Frank and Clementi, Cecilia}, - date = {2020-05-21}, - doi = {10.1063/5.0007276}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {19}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Ensemble Learning of Coarse-Grained Molecular Dynamics Force Fields with a Kernel Approach}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {152} -} - -@article{wangExtendibleGraphneuralnetworkbasedApproach2021, - abstract = {An accurate force field is the key to the success of all molecular mechanics simulations on organic polymers and biomolecules. Accuracy beyond density functional theory is often needed to describe the intermolecular interactions, while most correlated wavefunction (CW) methods are prohibitively expensive for large molecules. Therefore, it posts a great challenge to develop an extendible ab initio force field for large flexible organic molecules at CW level of accuracy. In this work, we face this challenge by combining the physics-driven nonbonding potential with a data-driven subgraph neural network bonding model (named sGNN). Tests on polyethylene glycol polymer chains show that our strategy is highly accurate and robust for molecules of different sizes. Therefore, we can develop the force field from small molecular fragments (with sizes easily accessible to CW methods) and safely transfer it to large polymers, thus opening a new path to the next-generation organic force fields.}, - archiveprefix = {arXiv}, - arxivid = {2106.00927}, - author = {Wang, Xufei and Xu, Yuanda and Zheng, Han and Yu, Kuang}, - date = {2021-06}, - eprint = {2106.00927}, - eprinttype = {arxiv}, - journaltitle = {arxiv.org}, - title = {An Extendible, Graph-Neural-Network-Based Approach for Accurate Force Field Development of Large Flexible Organic Molecules}, - url = {https://arxiv.org/abs/2106.00927} -} - -@article{wangMachineLearningCoarseGrained2019, - abstract = {Atomistic or ab initio molecular dynamics simulations are widely used to predict thermodynamics and kinetics and relate them to molecular structure. A common approach to go beyond the time- and length-scales accessible with such computationally expensive simulations is the definition of coarse-grained molecular models. Existing coarse-graining approaches define an effective interaction potential to match defined properties of high-resolution models or experimental data. In this paper, we reformulate coarse-graining as a supervised machine learning problem. We use statistical learning theory to decompose the coarse-graining error and cross-validation to select and compare the performance of different models. We introduce CGnets, a deep learning approach, that learns coarse-grained free energy functions and can be trained by a force-matching scheme. CGnets maintain all physically relevant invariances and allow one to incorporate prior physics knowledge to avoid sampling of unphysical structures. We show that CGnets can capture all-atom explicit-solvent free energy surfaces with models using only a few coarse-grained beads and no solvent, while classical coarse-graining methods fail to capture crucial features of the free energy surface. Thus, CGnets are able to capture multibody terms that emerge from the dimensionality reduction.}, - annotation = {WOS:000468624100008}, - author = {Wang, Jiang and Olsson, Simon and Wehmeyer, Christoph and Perez, Adria and Charron, Nicholas E. and de Fabritiis, Gianni and Noe, Frank and Clementi, Cecilia}, - date = {2019-05-22}, - doi = {10.1021/acscentsci.8b00913}, - issn = {2374-7943}, - journaltitle = {Acs Central Science}, - langid = {english}, - location = {{Washington}}, - number = {5}, - options = {useprefix=true}, - pages = {755--767}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {ACS Central Sci.}, - title = {Machine {{Learning}} of {{Coarse}}-{{Grained Molecular Dynamics Force Fields}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {5} -} - -@article{wangMultibodyEffectsCoarsegrained2021, - abstract = {The use of coarse-grained (CG) models is a popular approach to study complex biomolecular systems. By reducing the number of degrees of freedom, a CG model can explore long time- and length-scales inaccessible to computational models at higher resolution. If a CG model is designed by formally integrating out some of the system's degrees of freedom, one expects multi-body interactions to emerge in the effective CG model's energy function. In practice, it has been shown that the inclusion of multi-body terms indeed improves the accuracy of a CG model. However, no general approach has been proposed to systematically construct a CG effective energy that includes arbitrary orders of multi-body terms. In this work, we propose a neural network based approach to address this point and construct a CG model as a multi-body expansion. By applying this approach to a small protein, we evaluate the relative importance of the different multi-body terms in the definition of an accurate model. We observe a slow convergence in the multi-body expansion, where up to five-body interactions are needed to reproduce the free energy of an atomistic model.}, - author = {Wang, Jiang and Charron, Nicholas and Husic, Brooke and Olsson, Simon and Noé, Frank and Clementi, Cecilia}, - date = {2021-04}, - doi = {10.1063/5.0041022}, - journaltitle = {Journal of Chemical Physics}, - number = {16}, - publisher = {{American Institute of Physics Inc.}}, - title = {Multi-Body Effects in a Coarse-Grained Protein Force Field}, - url = {https://aip.scitation.org/doi/abs/10.1063/5.0041022}, - volume = {154} -} - -@article{wangPredictingAdsorptionAbility2021, - abstract = {Accurately evaluating the adsorption ability of adsorbents for heavy metal ions (HMIs) and organic pollutants in water is critical for the design and preparation of emerging highly efficient adsorbents. However, predicting adsorption capabilities of adsorbents at arbitrary sites is challenging, with currently unavailable measuring technology for active sites and the corresponding activities. Here, we present an efficient artificial intelligence (AI) approach to predict the adsorption ability of adsorbents at arbitrary sites, as a case study of three HMIs (Pb(II), Hg(II), and Cd(II)) adsorbed on the surface of a representative two-dimensional graphitic-C3N4. We apply the deep neural network and transfer learning to predict the adsorption capabilities of three HMIs at arbitrary sites, with the predicted results of Cd(II){$>$}Hg(II){$>$}Pb(II) and the root-mean-squared errors less than 0.1eV. The proposed AI method has the same prediction accuracy as the ab initio DFT calculation, but is millions of times faster than the DFT to predict adsorption abilities at arbitrary sites and only requires one-tenth of datasets compared to training from scratch. We further verify the adsorption capacity of g-C3N4 towards HMIs experimentally and obtain results consistent with the AI prediction. It indicates that the presented approach is capable of evaluating the adsorption ability of adsorbents efficiently, and can be further extended to other interdisciplines and industries for the adsorption of harmful elements in aqueous solution.}, - annotation = {WOS:000616403300003}, - author = {Wang, Zhilong and Zhang, Haikuo and Ren, Jiahao and Lin, Xirong and Han, Tianli and Liu, Jinyun and Li, Jinjin}, - date = {2021-01-29}, - doi = {10.1038/s41524-021-00494-9}, - journaltitle = {Npj Computational Materials}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - pages = {19}, - publisher = {{Nature Research}}, - shortjournal = {npj Comput. Mater.}, - title = {Predicting Adsorption Ability of Adsorbents at Arbitrary Sites for Pollutants Using Deep Transfer Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {7} -} - -@article{wangSymmetryadaptedGraphNeural2021, - abstract = {Molecular dynamics is a powerful simulation tool to explore material properties. Most of the realistic material systems are too large to be simulated with first-principles molecular dynamics. Classical molecular dynamics has lower computational cost but requires accurate force fields to achieve chemical accuracy. In this work, we develop a symmetry-adapted graph neural networks framework, named molecular dynamics graph neural networks (MDGNN), to construct force fields automatically for molecular dynamics simulations for both molecules and crystals. This architecture consistently preserves the translation, rotation and permutation invariance in the simulations. We propose a new feature engineering method including higher order contributions and show that MDGNN accurately reproduces the results of both classical and first-principles molecular dynamics. We also demonstrate that force fields constructed by the model has good transferability. Therefore, MDGNN provides an efficient and promising option for molecular dynamics simulations of large scale systems with high accuracy.}, - archiveprefix = {arXiv}, - author = {Wang, Zun and Wang, Chong and Zhao, Sibo and Du, Shiqiao and Xu, Yong and Gu, Bing-Lin and Duan, Wenhui}, - date = {2021}, - eprint = {2101.02930}, - eprinttype = {arxiv}, - title = {Symmetry-Adapted Graph Neural Networks for Constructing Molecular Dynamics Force Fields}, - url = {https://arxiv.org/abs/2101.02930} -} - -@article{Weinan2020, - abstract = {Machine learning is poised as a very powerful tool that can drastically improve our ability to carry out scientific research. However, many issues need to be addressed before this becomes a reality. This article focuses on one particular issue of broad interest: How can we integrate machine learning with physics-based modeling to develop new interpretable and truly reliable physical models? After introducing the general guidelines, we discuss the two most important issues for developing machine learning-based physical models: Imposing physical constraints and obtaining optimal datasets. We also provide a simple and intuitive explanation for the fundamental reasons behind the success of modern machine learning, as well as an introduction to the concurrent machine learning framework needed for integrating machine learning with physics-based modeling. Molecular dynamics and moment closure of kinetic equations are used as examples to illustrate the main issues discussed. We end with a general discussion on where this integration will lead us to, and where the new frontier will be after machine learning is successfully integrated into scientific modeling.}, - author = {Weinan, E and Han, Jiequn and Linfeng, Zhang}, - date = {2020-06}, - title = {Integrating Machine Learning with Physics-Based Modeling}, - url = {https://arxiv.org/abs/2006.02619 http://arxiv.org/abs/2006.02619 https://deepai.org/publication/integrating-machine-learning-with-physics-based-modeling} -} - -@article{weinreichPropertiesAlphaBrassNanoparticles2020, - abstract = {Binary metal clusters are of high interest for applications in heterogeneous catalysis and have received much attention in recent years. To gain insights into their structure and composition at the atomic scale, computer simulations can provide valuable information if reliable interatomic potentials are available. In this paper we describe the construction of a high-dimensional neural network potential (HDNNP) intended for simulations of large brass nanoparticles with thousands of atoms, which is also applicable to bulk a-brass and its surfaces. The HDNNP, which is based on reference data obtained from density-functional theory calculations, is very accurate with a root-mean-square error of 1.7 meV/atom for total energies and 39 meV angstrom(-1) for the forces of structures not included in the training set. The potential has been thoroughly validated for a wide range of energetic and structural properties of bulk a-brass, its surfaces as well as clusters of different size and composition demonstrating its suitability for large-scale molecular dynamics and Monte Carlo simulations with firstprinciples accuracy.}, - annotation = {WOS:000541745800055}, - author = {Weinreich, Jan and Roemer, Anton and Paleico, Martin Leandro and Behler, Joerg}, - date = {2020-06-11}, - doi = {10.1021/acs.jpcc.0c00559}, - issn = {1932-7447}, - journaltitle = {Journal of Physical Chemistry C}, - langid = {english}, - location = {{Washington}}, - number = {23}, - pages = {12682--12695}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. C}, - title = {Properties of Alpha-{{Brass Nanoparticles}}. 1. {{Neural Network Potential Energy Surface}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {124} -} - -@article{wenDevelopmentDeepMachine2019, - abstract = {Interatomic potentials based on neural-network machine learning (ML) approach to address the long-standing challenge of accuracy versus efficiency in molecular-dynamics simulations have recently attracted a great deal of interest. Here, utilizing Pd-Si system as a prototype, we extend the development of neural-network ML potentials to compounds exhibiting various types of bonding characteristics. The ML potential is trained by fitting to the energies and forces of both liquid and crystal structures first-principles calculations based on density-functional theory (DFT). We show that the generated ML potential captures the structural features and motifs in Pd82Si18 and Pd75Si25 liquids more accurately than the existing interatomic potential based on embedded-atom method (EAM). The ML potential also describes the solid-liquid interface of these systems very well. Moreover, while the existing EAM potential fails to describe the relative energies of various crystalline structures and predict wrong ground-state structures at Pd3Si and Pd9Si2 composition, the developed ML potential predicts correctly the ground-state structures from genetic algorithm search. The efficient ML potential with DFT accuracy from our study will provide a promising scheme for accurate atomistic simulations of structures and dynamics of complex Pd-Si system.}, - annotation = {WOS:000494024200001}, - author = {Wen, Tongqi and Wang, Cai-Zhuang and Kramer, M. J. and Sun, Yang and Ye, Beilin and Wang, Haidi and Liu, Xueyuan and Zhang, Chao and Zhang, Feng and Ho, Kai-Ming and Wang, Nan}, - date = {2019-11-04}, - doi = {10.1103/PhysRevB.100.174101}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {17}, - pages = {174101}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Development of a Deep Machine Learning Interatomic Potential for Metalloid-Containing {{Pd}}-{{Si}} Compounds}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {100} -} - -@article{westermayrCombiningSchNetSHARC2020, - abstract = {In recent years, deep learning has become a part of our everyday life and is revolutionizing quantum chemistry as well. In this work, we show how deep learning can be used to advance the research field of photochemistry by learning all important properties-multiple energies, forces, and different couplings-for photodynamics simulations. We simplify such simulations substantially by (i) a phase-free training skipping costly preprocessing of raw quantum chemistry data; (ii) rotationally covariant nonadiabatic couplings, which can either be trained or (iii) alternatively be approximated from only ML potentials, their gradients, and Hessians; and (iv) incorporating spin-orbit couplings. As the deep-learning method, we employ SchNet with its automatically determined representation of molecular structures and extend it for multiple electronic states. In combination with the molecular dynamics program SHARC, our approach termed SchNarc is tested on two polyatomic molecules and paves the way toward efficient photodynamics simulations of complex systems.}, - annotation = {WOS:000537432500012}, - author = {Westermayr, Julia and Gastegger, Michael and Marquetand, Philipp}, - date = {2020-05-21}, - doi = {10.1021/acs.jpclett.0c00527}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {10}, - pages = {3828--3834}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - shorttitle = {Combining {{SchNet}} and {{SHARC}}}, - title = {Combining {{SchNet}} and {{SHARC}}: {{The SchNarc Machine Learning Approach}} for {{Excited}}-{{State Dynamics}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{westermayrMachineLearningExcitedstate2020, - abstract = {Machine learning is employed at an increasing rate in the research field of quantum chemistry. While the majority of approaches target the investigation of chemical systems in their electronic ground state, the inclusion of light into the processes leads to electronically excited states and gives rise to several new challenges. Here, we survey recent advances for excited-state dynamics based on machine learning. In doing so, we highlight successes, pitfalls, challenges and future avenues for machine learning approaches for light-induced molecular processes.}, - author = {Westermayr, Julia and Marquetand, Philipp}, - date = {2020-09-19}, - doi = {10/gksxpp}, - issn = {2632-2153}, - journaltitle = {Machine Learning: Science and Technology}, - langid = {english}, - number = {4}, - pages = {043001}, - shortjournal = {Mach. Learn.: Sci. Technol.}, - title = {Machine Learning and Excited-State Molecular Dynamics}, - url = {https://iopscience.iop.org/article/10.1088/2632-2153/ab9c3e}, - urldate = {2021-08-11}, - volume = {1} -} - -@article{willattAtomdensityRepresentationsMachine2019a, - abstract = {The applications of machine learning techniques to chemistry and materials science become more numerous by the day. The main challenge is to devise representations of atomic systems that are at the same complete and concise, so as to reduce the number of reference calculations that are needed to predict the properties of different types of materials reliably. This has led to a proliferation of alternative ways to convert an atomic structure into an input for a machine-learning model. We introduce an abstract definition of chemical environments that is based on a smoothed atomic density, using a bra-ket notation to emphasize basis set independence and to highlight the connections with some popular choices of representations for describing atomic systems. The correlations between the spatial distribution of atoms and their chemical identities are computed as inner products between these feature kets, which can be given an explicit representation in terms of the expansion of the atom density on orthogonal basis functions, that is equivalent to the smooth overlap of atomic positions power spectrum, but also in real space, corresponding to n-body correlations of the atom density. This formalism lays the foundations for a more systematic tuning of the behavior of the representations, by introducing operators that represent the correlations between structure, composition, and the target properties. It provides a unifying picture of recent developments in the field and indicates a way forward toward more effective and computationally affordable machine-learning schemes for molecules and materials. Published under license by AIP Publishing.}, - annotation = {WOS:000465442100014}, - author = {Willatt, Michael J. and Musil, Flix and Ceriotti, Michele}, - date = {2019-04-21}, - doi = {10.1063/1.5090481}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {15}, - pages = {154110}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Atom-Density Representations for Machine Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {150} -} - -@article{willattFeatureOptimizationAtomistic2018, - abstract = {Machine-learning of atomic-scale properties amounts to extracting correlations between structure, composition and the quantity that one wants to predict. Representing the input structure in a way that best reflects such correlations makes it possible to improve the accuracy of the model for a given amount of reference data. When using a description of the structures that is transparent and well-principled, optimizing the representation might reveal insights into the chemistry of the data set. Here we show how one can generalize the SOAP kernel to introduce a distance-dependent weight that accounts for the multi-scale nature of the interactions, and a description of correlations between chemical species. We show that this improves substantially the performance of ML models of molecular and materials stability, while making it easier to work with complex, multi-component systems and to extend SOAP to coarse-grained intermolecular potentials. The element correlations that give the best performing model show striking similarities with the conventional periodic table of the elements, providing an inspiring example of how machine learning can rediscover, and generalize, intuitive concepts that constitute the foundations of chemistry.}, - archiveprefix = {arXiv}, - author = {Willatt, Michael J. and Musil, Félix and Ceriotti, Michele}, - date = {2018}, - doi = {10/gfz26d}, - eprint = {1807.00236}, - eprinttype = {arxiv}, - issn = {1463-9076, 1463-9084}, - journaltitle = {Physical Chemistry Chemical Physics}, - langid = {english}, - note = {Comment: 9 pages, 4 figures}, - number = {47}, - pages = {29661--29668}, - shortjournal = {Phys. Chem. Chem. Phys.}, - title = {Feature {{Optimization}} for {{Atomistic Machine Learning Yields A Data}}-{{Driven Construction}} of the {{Periodic Table}} of the {{Elements}}}, - url = {http://arxiv.org/abs/1807.00236}, - urldate = {2021-08-11}, - volume = {20} -} - -@article{wirnsbergerTargetedFreeEnergy2020a, - abstract = {Free energy perturbation (FEP) was proposed by Zwanzig [J. Chem. Phys. 22, 1420 (1954)] more than six decades ago as a method to estimate free energy differences and has since inspired a huge body of related methods that use it as an integral building block. Being an importance sampling based estimator, however, FEP suffers from a severe limitation: the requirement of sufficient overlap between distributions. One strategy to mitigate this problem, called Targeted FEP, uses a high-dimensional mapping in configuration space to increase the overlap of the underlying distributions. Despite its potential, this method has attracted only limited attention due to the formidable challenge of formulating a tractable mapping. Here, we cast Targeted FEP as a machine learning problem in which the mapping is parameterized as a neural network that is optimized so as to increase the overlap. We develop a new model architecture that respects permutational and periodic symmetries often encountered in atomistic simulations and test our method on a fully periodic solvation system. We demonstrate that our method leads to a substantial variance reduction in free energy estimates when compared against baselines, without requiring any additional data.}, - annotation = {WOS:000584689800003}, - author = {Wirnsberger, Peter and Ballard, Andrew J. and Papamakarios, George and Abercrombie, Stuart and Racaniere, Sebastien and Pritzel, Alexander and Rezende, Danilo Jimenez and Blundell, Charles}, - date = {2020-10-14}, - doi = {10.1063/5.0018903}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {14}, - pages = {144112}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Targeted Free Energy Estimation via Learned Mappings}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {153} -} - -@article{Wu, - abstract = {In this work we present a general machine learning based scheme to optimize experimental control. The method utilizes the neural network to learn the relation between the control parameters and the control goal, with which the optimal control parameters can be obtained. The main challenge of this approach is that the labeled data obtained from experiments are not abundant. The central idea of our scheme is to use the active learning to overcome this difficulty. As a demonstration example, we apply our method to control evaporative cooling experiments in cold atoms. We have first tested our method with simulated data and then applied our method to real experiments. We demonstrate that our method can successfully reach the best performance within hundreds of experimental runs. Our method does not require knowledge of the experimental system as a prior and is universal for experimental control in different systems.}, - archiveprefix = {arXiv}, - arxivid = {2003.11804v2}, - author = {Wu, Y and Meng, Z and Wen, K and Mi, C and J, Zhang}, - eprint = {2003.11804v2}, - eprinttype = {arxiv}, - journaltitle = {iopscience.iop.org}, - title = {Active Learning Approach to Optimization of Experimental Control}, - url = {https://iopscience.iop.org/article/10.1088/0256-307X/37/10/103201/meta} -} - -@article{wuDeepLearningAccurate2021, - abstract = {The discovery of ferroelectricity in HfO2-based thin films opens up new opportunities for using this silicon-compatible ferroelectric to realize low-power logic circuits and high-density nonvolatile memories. The functional performances of ferroelectrics are intimately related to their dynamic responses to external stimuli such as electric fields at finite temperatures. Molecular dynamics is an ideal technique for investigating dynamical processes on large length and time scales, though its applications to new materials are often hindered by the limited availability and accuracy of classical force fields. Here we present a deep neural network-based interatomic force field of HfO2 learned from ab initio data using a concurrent learning procedure. The model potential is able to predict structural properties such as elastic constants, equation of states, phonon dispersion relationships, and phase transition barriers of various hafnia polymorphs with accuracy comparable with density functional theory calculations. The validity of this model potential is further confirmed by the reproduction of experimental sequences of temperature-driven ferroelectric-paraelectric phase transitions of HfO2 with isobaric-isothermal ensemble molecular dynamics simulations. We suggest a general approach to extend the model potential of HfO(2 )to related material systems including dopants and defects.}, - annotation = {WOS:000613140900002}, - author = {Wu, Jing and Zhang, Yuzhi and Zhang, Linfeng and Liu, Shi}, - date = {2021-01-29}, - doi = {10.1103/PhysRevB.103.024108}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {2}, - pages = {024108}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Deep Learning of Accurate Force Field of Ferroelectric {{HfO2}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {103} -} - -@article{wuDeepLearningAccurate2021, - abstract = {The discovery of ferroelectricity in HfO2-based thin films opens up new opportunities for using this silicon-compatible ferroelectric to realize low-power logic circuits and high-density nonvolatile memories. The functional performances of ferroelectrics are intimately related to their dynamic responses to external stimuli such as electric fields at finite temperatures. Molecular dynamics is an ideal technique for investigating dynamical processes on large length and time scales, though its applications to new materials are often hindered by the limited availability and accuracy of classical force fields. Here we present a deep neural network-based interatomic force field of HfO2 learned from ab initio data using a concurrent learning procedure. The model potential is able to predict structural properties such as elastic constants, equation of states, phonon dispersion relationships, and phase transition barriers of various hafnia polymorphs with accuracy comparable with density functional theory calculations. The validity of this model potential is further confirmed by the reproduction of experimental sequences of temperature-driven ferroelectric-paraelectric phase transitions of HfO2 with isobaric-isothermal ensemble molecular dynamics simulations. We suggest a general approach to extend the model potential of HfO(2 )to related material systems including dopants and defects.}, - annotation = {WOS:000613140900002}, - author = {Wu, Jing and Zhang, Yuzhi and Zhang, Linfeng and Liu, Shi}, - date = {2021-01-29}, - doi = {10.1103/PhysRevB.103.024108}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {2}, - pages = {024108}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Deep Learning of Accurate Force Field of Ferroelectric {{HfO2}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {103} -} - -@article{wuModelingMetalNanoparticles2020, - abstract = {Interatomic potential plays a key role in ensuring the accuracy and reliability of molecular-dynamics simulation results. While most empirical potentials are benchmarked against a set of carefully chosen bulk material properties, recent advances in machine learning have seen the emergence of neural-network-based mathematical potentials capable of describing highly complex potential energy surfaces for a variety of systems. We report here the development of a neural-network interatomic potential (NNIP) with modified embedded-atom method background density as fingerprint functions, which could accurately model the energetics of metallic nanoparticles and clusters (Cu as a representative example) widely used in catalysis. To appropriately account for the diverse chemical environments encountered in nanoparticles/nanoclusters, an extensive set of atomic configurations (totaling 18 084) were calculated using density-functional-theory (DFT) at the Perdew-Burke-Ernzerhof level. In addition to standard bulk properties such as cohesive energies and elastic constants, the sampled configurations also include a substantial number of differently oriented crystal facets and differently sized nanoparticles and nanoclusters, greatly expanding the value range of NNIP features that was otherwise quite limited. The complex energy potential surface of Cu can be faithfully reproduced, with an average error of 0.011 eV/at for energy states within 3 eV of the ground state. As an illustration, the developed NNIP is used to simulate the molecular dynamics of copper nanoparticles, and good agreement is achieved between DFT and the NNIP.}, - annotation = {WOS:000577208700002}, - author = {Wu, Feifeng and Min, Hang and Wen, Yanwei and Chen, Rong and Zhao, Yunkun and Ford, Mike and Shan, Bin}, - date = {2020-10-15}, - doi = {10.1103/PhysRevB.102.144107}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {14}, - pages = {144107}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - shorttitle = {Modeling of Metal Nanoparticles}, - title = {Modeling of Metal Nanoparticles: {{Development}} of Neural-Network Interatomic Potential Inspired by Features of the Modified Embedded-Atom Method}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {102} -} - -@article{xiaHighThroughputStudyLattice2020, - abstract = {Thermal transport phenomena are ubiquitous and play a critical role in the performance of various microelectronic and energy-conversion devices. Binary rocksalt and zinc blende compounds, despite their rather simple crystal structures, exhibit an extraordinary range of lattice thermal conductivity (kappa(L)) spanning over 3 orders of magnitude. A comprehensive understanding of the underlying heat transfer mechanism through the development of microscopic theories is therefore of fundamental importance, yet it remains elusive because of the challenges arising from explicitly treating higher-order anharmonicity. Recent theoretical and experimental advances have revealed the essential role of quartic anharmonicity in suppressing heat transfer in zinc blende boron arsenide (BAs) with ultrahigh kappa(L). However, critical questions concerning the general effects of higher-order anharmonicity in the broad classes and chemistries of binary solids are still unanswered. Using our recently developed high-throughput phonon framework based on first-principles density functional theory calculations, we systematically investigate the lattice dynamics and thermal transport properties of 37 binary compounds with rocksalt and zinc blende structures at room temperature, with a particular focus on unraveling the impacts of quartic anharmonicity on kappa(L). Our advanced theoretical model for computing kappa(L) embraces current state-of-the-art methods, featuring a complete treatment of quartic anharmonicity for both phonon frequencies and lifetimes at finite temperatures, as well as contributions from off-diagonal terms in the heat-flux operator. We find the impacts of quartic anharmonicity on kappa(L) to be strikingly different in rocksalt and zinc blende compounds, owing to the countervailing effects on finite-temperature-induced shifts in phonon frequencies and scattering rates. By correlating.L with the phonon scattering phase space, we outline a qualitative but efficient route to assess the importance of four-phonon scattering from harmonic phonon calculations. Among notable examples, in zinc blende HgTe, we identify an unprecedented sixfold reduction in kappa(L) due to four-phonon scattering, which dominates over the three-phonon scattering in the acoustic region at room temperature. We also demonstrate a possible breakdown of the phonon gas model in rocksalt AgCl, wherein the phonon states are significantly broadened due to strong intrinsic anharmonicity, inducing off-diagonal contributions to kappa(L) comparable to the diagonal ones. The deep physical insights gained in this work can be used to guide the rational design of thermal management materials.}, - annotation = {WOS:000588236900001}, - author = {Xia, Yi and Hegde, Vinay and Pal, Koushik and Hua, Xia and Gaines, Dale and Patel, Shane and He, Jiangang and Aykol, Muratahan and Wolverton, Chris}, - date = {2020-11-10}, - doi = {10.1103/PhysRevX.10.041029}, - issn = {2160-3308}, - journaltitle = {Physical Review X}, - langid = {english}, - location = {{College Pk}}, - number = {4}, - pages = {041029}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. X}, - title = {High-{{Throughput Study}} of {{Lattice Thermal Conductivity}} in {{Binary Rocksalt}} and {{Zinc Blende Compounds Including Higher}}-{{Order Anharmonicity}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {10} -} - -@article{Xie2021, - abstract = {We present a variational density matrix approach to the thermal properties of interacting fermions in the continuum. The variational density matrix is parametrized by a permutation equivariant many-body unitary transformation together with a discrete probabilistic model. The unitary transformation is implemented as a quantum counterpart of neural canonical transformation, which incorporates correlation effects via a flow of fermion coordinates. As the first application, we study electrons in a two-dimensional quantum dot with an interaction-induced crossover from Fermi liquid to Wigner molecule. The present approach provides accurate results in the low-temperature regime, where conventional quantum Monte Carlo methods face severe difficulties due to the fermion sign problem. The approach is general and flexible for further extensions, thus holds the promise to deliver new physical results on strongly correlated fermions in the context of ultracold quantum gases, condensed matter, and warm dense matter physics.}, - author = {Xie, Hao and Zhang, Linfeng and Wang, Lei}, - date = {2021-05}, - journaltitle = {arxiv.org}, - title = {Ab-Initio Study of Interacting Fermions at Finite Temperature with Neural Canonical Transformation}, - url = {https://arxiv.org/abs/2105.08644} -} - -@article{xieBayesianForceFields2021a, - abstract = {We present a way to dramatically accelerate Gaussian process models for interatomic force fields based on many-body kernels by mapping both forces and uncertainties onto functions of low-dimensional features. This allows for automated active learning of models combining near-quantum accuracy, built-in uncertainty, and constant cost of evaluation that is comparable to classical analytical models, capable of simulating millions of atoms. Using this approach, we perform large-scale molecular dynamics simulations of the stability of the stanene monolayer. We discover an unusual phase transformation mechanism of 2D stanene, where ripples lead to nucleation of bilayer defects, densification into a disordered multilayer structure, followed by formation of bulk liquid at high temperature or nucleation and growth of the 3D bcc crystal at low temperature. The presented method opens possibilities for rapid development of fast accurate uncertainty-aware models for simulating long-time large-scale dynamics of complex materials.}, - annotation = {WOS:000631155400002}, - author = {Xie, Yu and Vandermause, Jonathan and Sun, Lixin and Cepellotti, Andrea and Kozinsky, Boris}, - date = {2021-03-19}, - doi = {10.1038/s41524-021-00510-y}, - journaltitle = {Npj Computational Materials}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - pages = {40}, - publisher = {{Nature Research}}, - shortjournal = {npj Comput. Mater.}, - title = {Bayesian Force Fields from Active Learning for Simulation of Inter-Dimensional Transformation of Stanene}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {7} -} - -@article{xieGraphDynamicalNetworks2019a, - abstract = {Understanding the dynamical processes that govern the performance of functional materials is essential for the design of next generation materials to tackle global energy and environmental challenges. Many of these processes involve the dynamics of individual atoms or small molecules in condensed phases, e.g. lithium ions in electrolytes, water molecules in membranes, molten atoms at interfaces, etc., which are difficult to understand due to the complexity of local environments. In this work, we develop graph dynamical networks, an unsupervised learning approach for understanding atomic scale dynamics in arbitrary phases and environments from molecular dynamics simulations. We show that important dynamical information, which would be difficult to obtain otherwise, can be learned for various multi-component amorphous material systems. With the large amounts of molecular dynamics data generated every day in nearly every aspect of materials design, this approach provides a broadly applicable, automated tool to understand atomic scale dynamics in material systems.}, - annotation = {WOS:000471758500003}, - author = {Xie, Tian and France-Lanord, Arthur and Wang, Yanming and Shao-Horn, Yang and Grossman, Jeffrey C.}, - date = {2019-06-17}, - doi = {10.1038/s41467-019-10663-6}, - issn = {2041-1723}, - journaltitle = {Nature Communications}, - langid = {english}, - location = {{London}}, - pages = {2667}, - publisher = {{Nature Publishing Group}}, - shortjournal = {Nat. Commun.}, - title = {Graph Dynamical Networks for Unsupervised Learning of Atomic Scale Dynamics in Materials}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {10} -} - -@article{xuAutomatedConstructionNeural, - abstract = {In recent years, the use of deep learning (neural network) potential energy surface (NNPES) in molecular dynamics simulation has experienced explosive growth as it can be as accurate as quantum chemistry methods while being as efficient as classical mechanic methods. However, the development of NNPES is highly non-trivial. In particular, it has been troubling to construct a dataset that is as small as possible yet can cover the target chemical space. In this work, an ESOINN-DP method is developed, which has the enhanced self-organizing incremental neuralnetwork (ESOINN) and a newly proposed error indicator at its core. With ESOINN-DP, One can construct the NNPES with little human intervention, and this method ensures that the constructed reference dataset covers the target chemical space with minimum redundancy. The performance of the ESOINN-DP method has been well validated by developing neural network potential energy surfaces for water clusters and by de-redundancy of a sub-data set of the ANI1 database. We believe that the ESOINN-DP method provides a novelty idea for the construction of NNPES and especially, the reference datasets, and it can be used for MD simulations of various gas-phase and condensed-phase chemical systems.}, - author = {Xu, Mingyuan and Zhu, Tong and Zhang, John Z H}, - langid = {english}, - pages = {18}, - title = {Automated {{Construction}} of {{Neural Network Potential Energy Surface}}: {{The Enhanced Self}}-{{Organizing Incremental Neural Network Deep Potential Method}}} -} - -@article{xuAutomaticallyConstructedNeural2021a, - abstract = {The development of accurate and efficient potential energy functions for the molecular dynamics simulation of metalloproteins has long been a great challenge for the theoretical chemistry community. An artificial neural network provides the possibility to develop potential energy functions with both the efficiency of the classical force fields and the accuracy of the quantum chemical methods. In this work, neural network potentials were automatically constructed by using the ESOINN-DP method for typical zinc proteins. For the four most common zinc coordination modes in proteins, the potential energy, atomic forces, and atomic charges predicted by neural network models show great agreement with quantum mechanics calculations and the neural network potential can maintain the coordination geometry correctly. In addition, MD simulation and energy optimization with the neural network potential can be readily used for structural refinement. The neural network potential is not limited by the function form and complex parameterization process, and important quantum effects such as polarization and charge transfer can be accurately considered. The algorithm proposed in this work can also be directly applied to proteins containing other metal ions.}, - annotation = {WOS:000668723600001}, - author = {Xu, Mingyuan and Zhu, Tong and Zhang, John Z. H.}, - date = {2021-06-18}, - doi = {10.3389/fchem.2021.692200}, - issn = {2296-2646}, - journaltitle = {Frontiers in Chemistry}, - langid = {english}, - location = {{Lausanne}}, - pages = {692200}, - publisher = {{Frontiers Media Sa}}, - shortjournal = {Front. Chem.}, - title = {Automatically {{Constructed Neural Network Potentials}} for {{Molecular Dynamics Simulation}} of {{Zinc Proteins}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {9} -} - -@article{xuDeeplearningPotentialCrystalline2020, - abstract = {This work investigates the ability of the deep-learning potential (DP) to describe structural, dynamic and energetic properties of crystalline and amorphous Li-Si alloys. Li-Si systems play an important role in the development of high-energy lithium ion batteries. One challenge in simulating Li-Si systems is to balance the proper description of complex Li-Si interactions and the system size. Molecular simulations implemented with DP provide a promising alternative to achieve this balance and enable us to investigate the fine details of Li-Si systems that the classical force fields cannot offer. We develop a DP for Li-Si systems with Li/Si ratio ranging from 0 to 4.2 based on a vast data set generated using the quantum mechanical calculations in an active learning procedure. Then we investigate the structural and dynamic properties of several crystalline and amorphous Li-Si systems using this developed DP. The DP can predict bulk densities, the radial distribution functions, and diffusivity of Li in amorphous Li-Si systems with an accuracy close to quantum mechanical calculations with the benefit of 20 times faster speed than the ab initio molecular dynamics simulations. Several issues related to the development of DP are also discussed.}, - author = {Xu, Nan and Shi, Yao and He, Yi and Shao, Qing}, - date = {2020-07}, - doi = {10.1021/acs.jpcc.0c03333}, - journaltitle = {Journal of Physical Chemistry C}, - number = {30}, - pages = {16278--16288}, - publisher = {{American Chemical Society}}, - title = {A Deep-Learning Potential for Crystalline and Amorphous Li-Si Alloys}, - url = {https://pubs.acs.org/doi/abs/10.1021/acs.jpcc.0c03333}, - volume = {124} -} - -@article{xuInitioMolecularDynamics, - abstract = {Artificial neural network provides the possibility to develop molecular potentials with both the efficiency of the classical molecular mechanics and the accuracy of the quantum chemical methods. In this work, we developed ab initio based neural network potential (NN/MM-RESPMBG) to perform molecular dynamics study for metalloproteins. The interaction energy, atomic forces, and atomic charges of metal binding group in NN/MM-RESP-MBG are described by a neural network potential trained with energies and forces generated from density functional calculations. Here, we used our recently proposed E-SOI-HDNN model to achieve the automatic construction of reference dataset of metalloproteins and the active learning of neural network potential functions. The predicted energies and atomic forces from the NN potential show excellent agreement with the quantum chemistry calculations. Using this approach, we can perform long time AIMD simulations and structure refinement MD simulation for metalloproteins. In 1 ns AIMD simulation of four common coordination mode of zinc-containing metalloproteins, the statistical average structure is in good agreement with statistic value of PDB Bank database. The neural network approach used in this study can be applied to construct potentials to metalloproteinase catalysis, ligand binding and other important biochemical processes and its salient features can shed light on the development of more accurate molecular potentials for metal ions in other biomacromolecule system.}, - author = {Xu, Mingyuan and Zhu, Tong and Zhang, John Z H}, - langid = {english}, - pages = {27}, - title = {Ab {{Initio Molecular Dynamics Simulation}} of {{Zinc}} Metalloproteins with {{Enhanced Self}}-{{Organizing Incremental High Dimensional Neural Network}}} -} - -@article{xuIsotopeEffectsMolecular2020, - abstract = {Feynman path-integral deep potential molecular dynamics (PI-DPMD) calculations have been employed to study both light (H2O) and heavy water (D2O) within the isothermalisobaric ensemble. In particular, the deep neural network is trained based on ab initio data obtained from the strongly constrained and appropriately normed (SCAN) exchange-correlation functional. Because of the lighter mass of hydrogen than deuteron, the properties of light water are more influenced by nuclear quantum effect than those of heavy water. Clear isotope effects are observed and analyzed in terms of hydrogen-bond structure and electronic properties of water that are closely associated with experimental observables. The molecular structures of both liquid H2O and D2O agree well with the data extracted from scattering experiments. The delicate isotope effects on radial distribution functions and angular distribution functions are well reproduced as well. Our approach demonstrates that deep neural network combined with SCAN functional based ab initio molecular dynamics provides an accurate theoretical tool for modeling water and its isotope effects.}, - annotation = {WOS:000600833600001}, - author = {Xu, Jianhang and Zhang, Chunyi and Zhang, Linfeng and Chen, Mohan and Santra, Biswajit and Wu, Xifan}, - date = {2020-12-21}, - doi = {10.1103/PhysRevB.102.214113}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {21}, - pages = {214113}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Isotope Effects in Molecular Structures and Electronic Properties of Liquid Water via Deep Potential Molecular Dynamics Based on the {{SCAN}} Functional}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {102} -} - -@article{xuIsotopeEffectsMolecular2020, - abstract = {Feynman path-integral deep potential molecular dynamics (PI-DPMD) calculations have been employed to study both light (H2O) and heavy water (D2O) within the isothermalisobaric ensemble. In particular, the deep neural network is trained based on ab initio data obtained from the strongly constrained and appropriately normed (SCAN) exchange-correlation functional. Because of the lighter mass of hydrogen than deuteron, the properties of light water are more influenced by nuclear quantum effect than those of heavy water. Clear isotope effects are observed and analyzed in terms of hydrogen-bond structure and electronic properties of water that are closely associated with experimental observables. The molecular structures of both liquid H2O and D2O agree well with the data extracted from scattering experiments. The delicate isotope effects on radial distribution functions and angular distribution functions are well reproduced as well. Our approach demonstrates that deep neural network combined with SCAN functional based ab initio molecular dynamics provides an accurate theoretical tool for modeling water and its isotope effects.}, - annotation = {WOS:000600833600001}, - author = {Xu, Jianhang and Zhang, Chunyi and Zhang, Linfeng and Chen, Mohan and Santra, Biswajit and Wu, Xifan}, - date = {2020-12-21}, - doi = {10.1103/PhysRevB.102.214113}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {21}, - pages = {214113}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Isotope Effects in Molecular Structures and Electronic Properties of Liquid Water via Deep Potential Molecular Dynamics Based on the {{SCAN}} Functional}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {102} -} - -@article{xuMolecularDynamicsSimulation2019a, - abstract = {An artificial neural network provides the possibility to develop molecular potentials with both the efficiency of the classical molecular mechanics and the accuracy of the quantum chemical methods. Here, we develop an ab initio based neural network potential (NN/MM-RESP) to perform molecular dynamics study of zinc ion in liquid water. In this approach, the interaction energy, atomic forces, and atomic charges of zinc ion and water molecules' in the first solvent shell are described by a neural network potential trained with energies and forces generated from density functional calculations. The predicted energies and forces from the NN potential show excellent agreement with the quantum chemistry calculations. Using this approach, we carried out molecular dynamics simulation to study the hydration of zinc ion in water. The experimentally observed zinc-water radial distribution function, as well as the X-ray absorption near edge structure spectrum, is well-reproduced by the MD simulation. Comparison of the results with other theoretical calculations is provided, and important features of the present approach are discussed. The neural network approach used in this study can be applied to construct potentials to study solvation of other metal ions, and its salient features can shed light on the development of more accurate molecular potentials for metal ions in other environments such as proteins.}, - annotation = {WOS:000486361700021}, - author = {Xu, Mingyuan and Zhu, Tong and Zhang, John Z. H.}, - date = {2019-08-01}, - doi = {10.1021/acs.jpca.9b04087}, - issn = {1089-5639}, - journaltitle = {Journal of Physical Chemistry A}, - langid = {english}, - location = {{Washington}}, - number = {30}, - pages = {6587--6595}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. A}, - title = {Molecular {{Dynamics Simulation}} of {{Zinc Ion}} in {{Water}} with an Ab {{Initio Based Neural Network Potential}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {123} -} - -@article{xuNovoMoleculeDesign, - abstract = {De novo molecule design through molecular generative model is gaining increasing attention in recent years. Here a novel generative model was proposed by integrating the 3D structural information of the protein binding pocket into the conditional RNN (cRNN) model to control the generation of drug-like molecules. In this model, the composition of protein binding pocket is effectively characterized through a coarse-grain strategy and the threedimensional information of the pocket can be represented by the sorted eigenvalues of the coulomb matrix (EGCM) of the coarse-grained atoms composing the binding pocket. In current work, we used our EGCM method and a previously reported binding pocket descriptor DeeplyTough to train cRNN models and compared their performance. It has been shown that the molecules generated with the control of protein environment information have a clear tendency on generating compounds with higher similarity to the original X-ray bound ligand than normal RNN model and also achieving better performance in terms of docking scores. Our results demonstrate the potential application of EGCM controlled generative model for the targeted molecule generation and guided exploration on the drug-like chemical space.}, - author = {Xu, Mingyuan and Ran, Ting and Chen, Hongming}, - langid = {english}, - pages = {25}, - title = {De Novo Molecule Design through Molecular Generative Model Conditioned by {{3D}} Information of Protein Binding Sites} -} - -@article{xuOptimizingTrainingData2021, - abstract = {Machine learning potential enables molecular dynamics simulations of systems beyond the capability of traditional force fields. One challenge in developing machine learning potential is how to construct a data set with low sample redundancy. This work investigates the method to optimize the training data set while maintaining the desirable accuracy of the machine learning potential using the structural similarity algorithm. We construct several subsets ranging from 200-1500 sample configurations by selecting representative configurations from a 6183-sample data set using the farthest point sampling method and examine the ability of the machine learning potential trained from the subsets to predict energy, atomic forces and structural properties of Li-Si systems. The simulation results show that the potential developed from 400 configurations can be as accurate as the one developed from the 6183-sample data set. In addition, our computation results highlight that the structure-comparison algorithms can not only effectively remove redundant from training sets, but also achieve an appropriate distribution of samples in training data sets.}, - archiveprefix = {arXiv}, - arxivid = {2103.04347}, - author = {Xu, Nan and Li, Chen and Shi, Yao and Shao, Qing and He, Yi}, - date = {2021-03}, - eprint = {2103.04347}, - eprinttype = {arxiv}, - journaltitle = {arxiv.org}, - title = {Optimizing Training Data Set for the Machine Learning Potential of Li-Si Alloys via Structural Similarity-Based Screening}, - url = {https://arxiv.org/abs/2103.04347 http://arxiv.org/abs/2103.04347} -} - -@article{xuPerspectiveComputationalReaction2021, - abstract = {Heterogeneous catalysis plays a significant role in the modern chemical industry. Towards the rational design of novel catalysts, understanding reactions over surfaces is the most essential aspect. Typical industrial catalytic processes such as syngas conversion and methane utilisation can generate a large reaction network comprising thousands of intermediates and reaction pairs. This complexity not only arises from the permutation of transformations between species but also from the extra reaction channels offered by distinct surface sites. Despite the success in investigating surface reactions at the atomic scale, the huge computational expense of ab initio methods hinders the exploration of such complicated reaction networks. With the proliferation of catalysis studies, machine learning as an emerging tool can take advantage of the accumulated reaction data to emulate the output of ab initio methods towards swift reaction prediction. Here, we briefly summarise the conventional workflow of reaction prediction, including reaction network generation, ab initio thermodynamics and microkinetic modelling. An overview of the frequently used regression models in machine learning is presented. As a promising alternative to full ab initio calculations, machine learning interatomic potentials are highlighted. Furthermore, we survey applications assisted by these methods for accelerating reaction prediction, exploring reaction networks, and computational catalyst design. Finally, we envisage future directions in computationally investigating reactions and implementing machine learning algorithms in heterogeneous catalysis.}, - annotation = {WOS:000648898600001}, - author = {Xu, Jiayan and Cao, Xiao-Ming and Hu, P.}, - date = {2021-05-21}, - doi = {10.1039/d1cp01349a}, - issn = {1463-9076}, - journaltitle = {Physical Chemistry Chemical Physics}, - langid = {english}, - location = {{Cambridge}}, - number = {19}, - pages = {11155--11179}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Phys. Chem. Chem. Phys.}, - title = {Perspective on Computational Reaction Prediction Using Machine Learning Methods in Heterogeneous Catalysis}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {23} -} - -@article{Yang, - abstract = {The study of chemical reactions in aqueous media is very important for its implications in several fields of science, from biology to industrial processes. However, modeling these reactions is difficult when water directly participates in the reaction, since it requires a fully quantum mechanical description of the system. Ab-initio molecular dynamics is the ideal candidate to shed light on these processes. However, its scope is limited by a high computational cost. A popular alternative is to perform molecular dynamics simulations powered by machine learning potentials, trained on an extensive set of quantum mechanical calculations. Doing so reliably for reactive processes is difficult because it requires including very many intermediate and transition state configurations. In this study we used an active learning procedure accelerated by enhanced sampling to harvest such structures and to build a neural-network potential to study the urea decomposition process in water. This allowed us to obtain the free energy profiles of this important reaction in a wide range of temperatures, to discover several novel metastable states, and improve the accuracy of the kinetic rates calculations. Furthermore, we found that the formation of the zwitterionic intermediate has the same probability of occurring via an acidic or a basic pathway, which could be the cause of the insensitivity of reaction rates to the solution pH.}, - author = {Yang, M and Bonati, L and Polino, D and M, Parrinello}, - date = {2021}, - journaltitle = {Elsevier}, - title = {Using Metadynamics to Build Neural Network Potentials for Reactive Events: The Case of Urea Decomposition in Water}, - url = {https://www.sciencedirect.com/science/article/pii/S092058612100136X} -} - -@report{yangConstructionNeuralNetwork2021, - abstract = {Abstract Classical potentials are widely used to describe protein physics, due to their simplicity and accuracy, but they are continuously challenged as real applications become more demanding with time. Deep neural networks could help generating alternative ways of describing protein physics. Here we propose an unsupervised learning method to derive a neural network energy function for proteins. The energy function is a probability density model learned from plenty of 3D local structures which have been extensively explored by evolution. We tested this model on a few applications (assessment of protein structures, protein dynamics and protein sequence design), showing that the neural network can correctly recognize patterns in protein structures. In other words, the neural network learned some aspects of protein physics from experimental data.}, - author = {Yang, Huan and Xiong, Zhaoping and Zonta, Francesco}, - date = {2021-04-27}, - doi = {10.1101/2021.04.26.441401}, - institution = {{Biophysics}}, - langid = {english}, - title = {Construction of a Neural Network Energy Function for Protein Physics}, - type = {preprint}, - url = {http://biorxiv.org/lookup/doi/10.1101/2021.04.26.441401}, - urldate = {2021-08-11} -} - -@article{yangRoleWaterReaction2019a, - abstract = {Asymmetric 1,3-dipolar cycloadditions of azomethine ylides with activated olefins are among the most important and versatile methods for the synthesis of enantioenriched pyrroline and pyrrolidine derivatives. Despite both theoretical and practical importance, the role of water molecules in the reactivity and endo/exo selectivity remains unclear. To explore how water accelerates the reactions and improves the endo/exo selectivity of the cycloadditions of 1,3-dipole phthalazinium-2-dicyanomethanide (1) and two dipolarophiles, an ab initio-quality neural network potential that overcomes the computational bottleneck of explicitly considering water molecules was used. It is demonstrated that not only the nature of both the dipolarophile and the 1,3-dipole, but also the solvent medium, can perturb or even alter the reaction mechanism. An extreme case was found for the reaction of 1,3-dipole 1 with methyl vinyl ketone, in which the reaction mechanism changes from a concerted to a stepwise mode on going from MeCN to H2O as solvent, with formation of a zwitterionic intermediate that is a very shallow minimum on the energy surface. Thus, high stereocontrol can still be expected despite the stepwise nature of the mechanism. The results indicate that water can induce global polarization along the reaction coordinate and highlight the role of microsolvation effects and bulk-phase effects in reproducing the experimentally observed aqueous acceleration and enhanced endo/exo selectivity.}, - annotation = {WOS:000474808200018}, - author = {Yang, Xin and Zou, Jun and Wang, Yifei and Xue, Ying and Yang, Shengyong}, - date = {2019-06-21}, - doi = {10.1002/chem.201900617}, - issn = {0947-6539}, - journaltitle = {Chemistry-a European Journal}, - langid = {english}, - location = {{Weinheim}}, - number = {35}, - pages = {8289--8303}, - publisher = {{Wiley-V C H Verlag Gmbh}}, - shortjournal = {Chem.-Eur. J.}, - title = {Role of {{Water}} in the {{Reaction Mechanism}} and Endo/Exo {{Selectivity}} of 1,3-{{Dipolar Cycloadditions Elucidated}} by {{Quantum Chemistry}} and {{Machine Learning}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {25} -} - -@article{Yao2020, - abstract = {In large-scale computations of physical problems, one often encounters the situation of having to determine a multidimensional function, which can be numerically costly when computing each point in this multidimensional space is already time-demanding. In the work, we propose that the active learning algorithm can speed up such calculations. The basic idea is to fit a multidimensional function by neural networks, and the key point is to make the query of labeled data more economical by using a strategy called "query by committee." We present the general protocol of this fitting scheme, as well as the procedure of how to further compute physical observables with the fitted functions. We show that this method can work well with two examples, which are the quantum three-body problem in atomic physics and the anomalous Hall conductivity in condensed matter physics, respectively. In these examples, we show that one reaches an accuracy of a few percent error in computing physical observables, all the while using fewer than 10\% of total data points compared with uniform sampling. With these two examples, we also visualize that by using the active learning algorithm, the required amount of data points are added mostly in the regime where the function varies most rapidly, which explains the mechanism for the efficiency of the algorithm. We expect broad applications of our method to various kinds of computational physical problems. Neural network (NN) based supervised learning methods have nowadays found broad applications in studying quantum physics in condensed matter materials and atomic, molecular and optical systems [1,2]. On the theoretical side, applications include finding orders and topological invariants in quantum phases [3-9], generating variational wave functions for quantum many-body states [10-14] and speeding up quantum Monte Carlo sampling [15,16]. On the experimental side, these methods can help optimizing experimental protocols [17,18] and analyzing experimental data [19-21]. Usually the supervised learning scheme requires a huge set of labeled data. However, in many physical applications, labeling data can be quite expensive in terms of computational resources. For instance, performing computation or experiments repeatedly can be time-and resources-demanding. Therefore, in many cases, labeled data are not abundant, which is a challenge that has prevented many applications. The active learning is a scheme to solve this problem [22]. It starts from training a NN with a small initial data set, and then actively queries the labeled data based on the prediction of the NN and iteratively improves the performance of the NN until the goal of the task is reached. With this approach, sampling the large parameter space can be made more efficiently, and the demand of labeled data is usually much less than normal supervised learning methods. Recently a few works have applied the active learning algorithm to determination of the interatomic potentials in quantum materials [23-25] and to optimal control in quantum experiments [26,27]. In these situations, labeled data have to be obtained either by ab initio calculation or by repeating experiments, which are both time consuming. In this work, we focus on a class of general and common task in computational physics that is to numerically determine a multidimensional function, say, F (α 1 , α 2 ,. .. , α n), where α i are parameters. Supposing that we uniformly discretize each parameter into L points, there are then L n data points in total that need to be calculated. In many cases, calculation of each point already takes quite some time, and thus the total computational cost can be massive. Nevertheless, for most functions, there are regions where the function varies smoothly and regions where the function varies rapidly. Ideally, the goal is to sample more points in the steep regions and fewer points in the smooth regions in order to efficiently obtain a good fitting in the entire parameter space. However, it seems to be a paradox because one does not know which regions the function varies more rapidly prior to computing the function. Here we show that this goal can be achieved by using the active learning algorithm and the "query by committee" strategy [28] where data points are added iteratively. Below we will first introduce the general protocol and then demonstrate the algorithm in two concrete problems: the quantum three-body problem and the anomalous Hall conductivity problem.}, - author = {Yao, J and Wu, Y and Koo, J and Yan, B and H, Zhai}, - date = {2020-03}, - doi = {10.1103/physrevresearch.2.013287}, - journaltitle = {APS}, - number = {1}, - pages = {13287}, - publisher = {{American Physical Society (APS)}}, - title = {Active Learning Algorithm for Computational Physics}, - url = {https://journals.aps.org/prresearch/abstract/10.1103/PhysRevResearch.2.013287}, - volume = {2} -} - -@article{yaoNuclearQuantumEffect2021, - abstract = {We report structural and dynamical properties of liquid water described by the random phase approximation (RPA) correlation together with the exact exchange energy (EXX) within density functional theory. By utilizing thermostated ring polymer molecular dynamics, we examine the nuclear quantum effects and their temperature dependence. We circumvent the computational limitation of performing direct firstprinciples molecular dynamics simulation at this high level of electronic structure theory by adapting an artificial neural network model. We show that the EXX+RPA level of theory accurately describes liquid water in terms of both dynamical and structural properties.}, - author = {Yao, Yi and Kanai, Yosuke}, - date = {2021-07-15}, - doi = {10/gk5v27}, - issn = {1948-7185, 1948-7185}, - journaltitle = {The Journal of Physical Chemistry Letters}, - langid = {english}, - number = {27}, - pages = {6354--6362}, - shortjournal = {J. Phys. Chem. Lett.}, - title = {Nuclear {{Quantum Effect}} and {{Its Temperature Dependence}} in {{Liquid Water}} from {{Random Phase Approximation}} via {{Artificial Neural Network}}}, - url = {https://pubs.acs.org/doi/10.1021/acs.jpclett.1c01566}, - urldate = {2021-08-10}, - volume = {12} -} - -@article{yooAtomicEnergyMapping2019a, - abstract = {We investigate the atomic energy mapping inferred by machine-learning potentials, in particular neural network potentials. We first show that the transferable atomic energy can be defined within the density functional theory, which means that the core of machine-learning potentials is to deduce a reference atomic-energy function from the given set of total energies. By utilizing invariant points in the feature space at which the atomic energy has a fixed reference value, we examine the atomic energy mapping of neural network potentials. Examples on Si consistently support that NNPs are capable of learning correct atomic energies. However, we also find that the neural network potential is vulnerable to 'ad hoc' mapping in which the total energy appears to be trained accurately while the atomic energy mapping is incorrect in spite of its capability. We show that the energy mapping can be improved by choosing the training set carefully and monitoring the atomic energy at the invariant points during the training procedure. The energy mapping in multicomponent systems is also discussed.}, - annotation = {WOS:000483590500005}, - author = {Yoo, Dongsun and Lee, Kyuhyun and Jeong, Wonseok and Lee, Dongheon and Watanabe, Satoshi and Han, Seungwu}, - date = {2019-09-03}, - doi = {10.1103/PhysRevMaterials.3.093802}, - issn = {2475-9953}, - journaltitle = {Physical Review Materials}, - langid = {english}, - location = {{College Pk}}, - number = {9}, - pages = {093802}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Mater.}, - title = {Atomic Energy Mapping of Neural Network Potential}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {3} -} - -@article{youngTransferableActivelearningStrategy2021, - abstract = {Predictive molecular simulations require fast, accurate and reactive interatomic potentials. Machine learning offers a promising approach to construct such potentials by fitting energies and forces to high-level quantum-mechanical data, but doing so typically requires considerable human intervention and data volume. Here we show that, by leveraging hierarchical and active learning, accurate Gaussian Approximation Potential (GAP) models can be developed for diverse chemical systems in an autonomous manner, requiring only hundreds to a few thousand energy and gradient evaluations on a reference potential-energy surface. The approach uses separate intra- and inter-molecular fits and employs a prospective error metric to assess the accuracy of the potentials. We demonstrate applications to a range of molecular systems with relevance to computational organic chemistry: ranging from bulk solvents, a solvated metal ion and a metallocage onwards to chemical reactivity, including a bifurcating Diels-Alder reaction in the gas phase and non-equilibrium dynamics (a model S(N)2 reaction) in explicit solvent. The method provides a route to routinely generating machine-learned force fields for reactive molecular systems.}, - annotation = {WOS:000674966100001}, - author = {Young, Tom A. and Johnston-Wood, Tristan and Deringer, Volker L. and Duarte, Fernanda}, - date = {2021}, - doi = {10.1039/d1sc01825f}, - issn = {2041-6520}, - journaltitle = {Chemical Science}, - langid = {english}, - location = {{Cambridge}}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Chem. Sci.}, - title = {A Transferable Active-Learning Strategy for Reactive Molecular Force Fields}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06} -} - -@article{yueWhenShortrangeAtomistic2021a, - abstract = {We explore the role of long-range interactions in atomistic machine-learning models by analyzing the effects on fitting accuracy, isolated cluster properties, and bulk thermodynamic properties. Such models have become increasingly popular in molecular simulations given their ability to learn highly complex and multi-dimensional interactions within a local environment; however, many of them fundamentally lack a description of explicit long-range interactions. In order to provide a well-defined benchmark system with precisely known pairwise interactions, we chose as the reference model a flexible version of the Extended Simple Point Charge (SPC/E) water model. Our analysis shows that while local representations are sufficient for predictions of the condensed liquid phase, the short-range nature of machine-learning models falls short in representing cluster and vapor phase properties. These findings provide an improved understanding of the role of long-range interactions in machine learning models and the regimes where they are necessary.}, - author = {Yue, Shuwen and Muniz, Maria Carolina and Calegari Andrade, Marcos F. and Zhang, Linfeng and Car, Roberto and Panagiotopoulos, Athanassios Z.}, - date = {2021-01-21}, - doi = {10/gkcq6f}, - issn = {0021-9606, 1089-7690}, - journaltitle = {The Journal of Chemical Physics}, - langid = {english}, - number = {3}, - pages = {034111}, - shortjournal = {J. Chem. Phys.}, - title = {When Do Short-Range Atomistic Machine-Learning Models Fall Short?}, - url = {http://aip.scitation.org/doi/10.1063/5.0031215}, - urldate = {2021-08-10}, - volume = {154} -} - -@article{Zeng2020, - abstract = {Reactive molecular dynamics (MD) simulation is a powerful tool to study the reaction mechanism of complex chemical systems. Central to the method is the potential energy surface (PES) that can describe the breaking and formation of chemical bonds. The development of PES of both accurate and efficent has attracted significant effort in the past two decades. Recently developed Deep Potential (DP) model has the promise to bring ab initio accuracy to large-scale reactive MD simulations. However, for complex chemical reaction processes like pyrolysis, it remains challenging to generate reliable DP models with an optimal training dataset. In this work, a dataset construction 1 scheme for such a purpose was established. The employment of a concurrent learning algorithm allows us to maximize the exploration of the chemical space while minimize the redundancy of the dataset. This greatly reduces the cost of computational resources required by ab initio calculations. Based on this method, we constructed a dataset for the pyrolysis of n-dodecane, which contains 35,496 structures. The reactive MD simulation with the DP model trained based on this dataset revealed the pyrolysis mechanism of n-dodecane in detail, and the simulation results are in good agreement with the experimental measurements. In addition, this dataset shows excellent trans-ferability to different long-chain alkanes. These results demonstrate the advantages of the proposed method for constructing training datasets for similar systems. 2}, - author = {Zeng, J and Zhang, L and Wang, H and Zhu, T}, - date = {2020}, - title = {Explore the Chemical Space of Linear Alkanes Pyrolysis via Deep Potential Generator}, - url = {https://chemrxiv.org/engage/chemrxiv/article-details/60c74fbc337d6c7aece281ae} -} - -@article{Zeng2021, - abstract = {We develop a new Deep Potential-Range Correction (DPRc) machine learning potential for combined quantum mechanical/molecular mechanical (QM/MM) simulations of chemical reactions in the condensed phase. The new range correction enables short-ranged QM/MM interactions to be tuned for higher accuracy, and the correction smoothly vanishes within a specified cutoff. We further develop an active learning procedure for robust neural network training. We test the DPRc model and training procedure against a series of 6 non-enzymatic phosphoryl transfer reactions in solution that are important in mechanistic studies of RNA-cleaving enzymes. Specifically, we apply DPRc corrections to a base QM model and test its ability to reproduce free energy profiles generated from a target QM model. We perform these * To whom correspondence should be addressed 1 comparisons using the MNDO/d and DFTB2 semiempirical models because they produce free energy profiles which differ significantly from each other, thereby providing us a rigorous stress test for the DPRc model and training procedure. The comparisons show that accurate reproduction of the free energy profiles requires correction of the QM/MM interactions out to 6 Å. We further find that the model's initial training benefits from generating data from temperature replica exchange simulations and including high-temperature configurations into the fitting procedure so the resulting models are trained to properly avoid high-energy regions. A single DPRc model was trained to reproduce 4 different reactions and yielded good agreement with the free energy profiles made from the target QM/MM simulations. The DPRc model was further demonstrated to be transferable to 2D free energy surfaces and 1D free energy profiles that were not explicitly considered in the training. Examination of the computational performance of the DPRc model showed that it was fairly slow when run on CPUs, but was sped up almost 100-fold when using an NVIDIA V100 GPUs, resulting in almost negligible overhead. The new DPRc model and training procedure provide a potentially powerful new tool for the creation of next-generation QM/MM potentials for a wide spectrum of free energy applications ranging from drug discovery to enzyme design. 2}, - author = {Zeng, J and Giese, TJ and Ekesan, Ş and York, DM}, - date = {2021}, - title = {Development of Range-Corrected Deep Learning Potentials for Fast, Accurate Quantum Mechanical/Molecular Mechanical Simulations of Chemical Reactions in Solution}, - url = {https://chemrxiv.org/engage/chemrxiv/article-details/60c7557dbdbb899a20a3a828} -} - -@article{zengComplexReactionProcesses2020, - abstract = {Combustion is a complex chemical system which involves thousands of chemical reactions and generates hundreds of molecular species and radicals during the process. In this work, a neural network-based molecular dynamics (MD) simulation is carried out to simulate the benchmark combustion of methane. During MD simulation, detailed reaction processes leading to the creation of specific molecular species including various intermediate radicals and the products are intimately revealed and characterized. Overall, a total of 798 different chemical reactions were recorded and some new chemical reaction pathways were discovered. We believe that the present work heralds the dawn of a new era in which neural network-based reactive MD simulation can be practically applied to simulating important complex reaction systems at ab initio level, which provides atomic-level understanding of chemical reaction processes as well as discovery of new reaction pathways at an unprecedented level of detail beyond what laboratory experiments could accomplish. Gaining insights into combustion processes is challenging due to the complex reactions involved. The present work proposes a neural network potential model trained to ab initio data that enables to simulate the combustion of methane by predicting reactants, products and reaction intermediates.}, - annotation = {WOS:000593975100015}, - author = {Zeng, Jinzhe and Cao, Liqun and Xu, Mingyuan and Zhu, Tong and Zhang, John Z. H.}, - date = {2020-11-11}, - doi = {10.1038/s41467-020-19497-z}, - issn = {2041-1723}, - journaltitle = {Nature Communications}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - pages = {5713}, - publisher = {{Nature Research}}, - shortjournal = {Nat. Commun.}, - title = {Complex Reaction Processes in Combustion Unraveled by Neural Network-Based Molecular Dynamics Simulation}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{zengComplexReactionProcesses2020, - abstract = {Combustion is a complex chemical system which involves thousands of chemical reactions and generates hundreds of molecular species and radicals during the process. In this work, a neural network-based molecular dynamics (MD) simulation is carried out to simulate the benchmark combustion of methane. During MD simulation, detailed reaction processes leading to the creation of specific molecular species including various intermediate radicals and the products are intimately revealed and characterized. Overall, a total of 798 different chemical reactions were recorded and some new chemical reaction pathways were discovered. We believe that the present work heralds the dawn of a new era in which neural network-based reactive MD simulation can be practically applied to simulating important complex reaction systems at ab initio level, which provides atomic-level understanding of chemical reaction processes as well as discovery of new reaction pathways at an unprecedented level of detail beyond what laboratory experiments could accomplish. Gaining insights into combustion processes is challenging due to the complex reactions involved. The present work proposes a neural network potential model trained to ab initio data that enables to simulate the combustion of methane by predicting reactants, products and reaction intermediates.}, - annotation = {WOS:000593975100015}, - author = {Zeng, Jinzhe and Cao, Liqun and Xu, Mingyuan and Zhu, Tong and Zhang, John Z. H.}, - date = {2020-11-11}, - doi = {10.1038/s41467-020-19497-z}, - issn = {2041-1723}, - journaltitle = {Nature Communications}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - pages = {5713}, - publisher = {{Nature Research}}, - shortjournal = {Nat. Commun.}, - title = {Complex Reaction Processes in Combustion Unraveled by Neural Network-Based Molecular Dynamics Simulation}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {11} -} - -@article{zengExploringChemicalSpace2021, - abstract = {Reactive molecular dynamics (MD) simulation is a powerful tool to study the reaction mechanism of complex chemical systems. Central to the method is the potential energy surface (PES) that can describe the breaking and formation of chemical bonds. The development of both accurate and efficient PES has attracted significant effort in the past 2 decades. A recently developed deep potential (DP) model has the promise to bring ab initio accuracy to large-scale reactive MD simulations. However, for complex chemical reaction processes like pyrolysis, it remains challenging to generate reliable DP models with an optimal training data set. In this work, a data set construction scheme for such a purpose was established. The employment of a concurrent learning algorithm allows us to maximize the exploration of the chemical space while minimizing the redundancy of the data set. This greatly reduces the cost of computational resources required for ab initio calculations. Based on this method, we constructed a data set for the pyrolysis of n-dodecane, which contains 35 496 structures. The reactive MD simulation with the DP model trained based on this data set revealed the pyrolysis mechanism of n-dodecane in detail, and the simulation results are in good agreement with the experimental measurements. In addition, this data set shows excellent transferability to different long-chain alkanes. These results demonstrate the advantages of the proposed method for constructing training data sets for similar systems.}, - annotation = {WOS:000609088800063}, - author = {Zeng, Jinzhe and Zhang, Linfeng and Wang, Han and Zhu, Tong}, - date = {2021-01-07}, - doi = {10.1021/acs.energyfuels.0c03211}, - issn = {0887-0624}, - journaltitle = {Energy \& Fuels}, - langid = {english}, - location = {{Washington}}, - number = {1}, - pages = {762--769}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {Energy Fuels}, - title = {Exploring the {{Chemical Space}} of {{Linear Alkane Pyrolysis}} via {{Deep Potential GENerator}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {35} -} - -@article{zengNeuralNetworkBased2019, - abstract = {Understanding and prediction of the chemical reactions are fundamental demanding in the study of many complex chemical systems. Reactive molecular dynamics (MD) simulation has been widely used for this purpose as it can offer atomic details and can help us better interpret chemical reaction mechanisms. In this study, two reference datasets were constructed and corresponding neural network (NN) potentials were trained based on them. For given large-scale reaction systems, the NN potentials can predict the potential energy and atomic forces of DFT precision, while it is orders of magnitude faster than the conventional DFT calculation. With these two models, reactive MD simulations were performed to explore the combustion mechanisms of hydrogen and methane. Benefit from the high efficiency of the NN model, nanosecond MD trajectories for large-scale systems containing hundreds of atoms were produced and detailed combustion mechanism was obtained. Through further development, the algorithms in this study can be used to explore and discovery reaction mechanisms of many complex reaction systems, such as combustion, synthesis, and heterogeneous catalysis without any predefined reaction coordinates and elementary reaction steps.}, - archiveprefix = {arXiv}, - arxivid = {1911.12252}, - author = {Zeng, Jinzhe and Cao, Liqun and Xu, Mingyuan and Zhu, Tong and Zhang, John ZH}, - date = {2019-11}, - eprint = {1911.12252}, - eprinttype = {arxiv}, - journaltitle = {arxiv.org}, - title = {Neural Network Based in Silico Simulation of Combustion Reactions}, - url = {http://arxiv.org/abs/1911.12252 https://arxiv.org/abs/1911.12252} -} - -@article{zepeda-nunezDeepDensityCircumventing2019, - abstract = {The recently developed Deep Potential [Phys. Rev. Lett. 120, 143001, 2018] is a powerful method to represent general inter-atomic potentials using deep neural networks. The success of Deep Potential rests on the proper treatment of locality and symmetry properties of each component of the network. In this paper, we leverage its network structure to effectively represent the mapping from the atomic configuration to the electron density in Kohn-Sham density function theory (KS-DFT). By directly targeting at the self-consistent electron density, we demonstrate that the adapted network architecture, called the Deep Density, can effectively represent the electron density as the linear combination of contributions from many local clusters. The network is constructed to satisfy the translation, rotation, and permutation symmetries, and is designed to be transferable to different system sizes. We demonstrate that using a relatively small number of training snapshots, Deep Density achieves excellent performance for one-dimensional insulating and metallic systems, as well as systems with mixed insulating and metallic characters. We also demonstrate its performance for real three-dimensional systems, including small organic molecules, as well as extended systems such as water (up to 512 molecules) and aluminum (up to 256 atoms).}, - archiveprefix = {arXiv}, - arxivid = {1912.00775v1}, - author = {Zepeda-Núñez, Leonardo and Chen, Yixiao and Zhang, Jiefu and Jia, Weile and Zhang, Linfeng and Lin, Lin}, - date = {2019}, - eprint = {1912.00775v1}, - eprinttype = {arxiv}, - journaltitle = {Elsevier}, - title = {Deep {{Density}}: Circumventing the {{Kohn}}-{{Sham}} Equations via Symmetry Preserving Neural Networks}, - url = {https://www.sciencedirect.com/science/article/pii/S0021999121004186} -} - -@article{zhaiActiveLearningManybody2020, - abstract = {The efficient selection of representative configurations that are used in high-level electronic structure calculations needed for the development of many-body molecular models poses a challenge to current data-driven approaches to molecular simulations. Here, we introduce an active learning (AL) framework for generating training sets corresponding to individual many-body contributions to the energy of an N-body system, which are required for the development of MB-nrg potential energy functions (PEFs). Our AL framework is based on uncertainty and error estimation and uses Gaussian process regression to identify the most relevant configurations that are needed for an accurate representation of the energy landscape of the molecular system under examination. Taking the Cs+-water system as a case study, we demonstrate that the application of our AL framework results in significantly smaller training sets than previously used in the development of the original MB-nrg PEF, without loss of accuracy. Considering the computational cost associated with high-level electronic structure calculations, our AL framework is particularly well-suited to the development of many-body PEFs, with chemical and spectroscopic accuracy, for molecular-level computer simulations from the gas to the condensed phase.}, - annotation = {WOS:000526884200001}, - author = {Zhai, Yaoguang and Caruso, Alessandro and Gao, Sicun and Paesani, Francesco}, - date = {2020-04-14}, - doi = {10.1063/5.0002162}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {14}, - pages = {144103}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - shorttitle = {Active Learning of Many-Body Configuration Space}, - title = {Active Learning of Many-Body Configuration Space: {{Application}} to the {{Cs}}+-Water {{MB}}-Nrg Potential Energy Function as a Case Study}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {152} -} - -@article{zhaiBubbleNetInferringMicrobubble2021, - abstract = {Micro-bubbles and bubbly flows are widely observed and applied in chemical engineering, medicine, involves deformation, rupture, and collision of bubbles, phase mixture, etc. We study bubble dynamics by setting up two numerical simulation cases: bubbly flow with a single bubble and multiple bubbles, both confined in the microchannel, with parameters corresponding to their medical backgrounds. Both the cases have their medical background applications. Multiphase flow simulation requires high computation accuracy due to possible component losses that may be caused by sparse meshing during the computation. Hence, data-driven methods can be adopted as an useful tool. Based on physics-informed neural networks (PINNs), we propose a novel deep learning framework BubbleNet, which entails three main parts: deep neural networks (DNN) with sub nets for predicting different physics fields; the semi-physics-informed part, with only the fluid continuum condition and the pressure Poisson equation P encoded within; the time discretized normalizer (TDN), an algorithm to normalize field data per time step before training. We apply the traditional DNN and our BubbleNet to train the coarsened simulation data and predict the physics fields of both the two bubbly flow cases. The BubbleNets are trained for both with and without P, from which we conclude that the ’physics-informed’ part can serve as an inner supervision. Results indicate our framework can predict the physics fields more accurately, estimating the prediction absolute errors. Our deep learning predictions outperforms traditional numerical methods computed with similar data density meshing. The proposed network can potentially be applied to many other engineering fields.}, - archiveprefix = {arXiv}, - author = {Zhai, Hanfeng and Hu, Guohui}, - date = {2021-07-26}, - eprint = {2105.07179}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: Edit and reconstruct the paper; remade some figures; \& reanalysis some results}, - primaryclass = {physics}, - shorttitle = {{{BubbleNet}}}, - title = {{{BubbleNet}}: {{Inferring}} Micro-Bubble Dynamics with Semi-Physics-Informed Deep Learning}, - url = {http://arxiv.org/abs/2105.07179}, - urldate = {2021-08-11} -} - -@article{Zhang2020c, - abstract = {In recent years, machine learning has emerged as a promising tool for dealing with the difficulty of representing high dimensional functions. This gives us an unprecedented opportunity to revisit theoretical foundations of various scientific fields, develop new schemes, improve existing methodologies, and solve problems that were too complicated for conventional approaches to address. In this dissertation, we identify a list of such problems in the context of multiscale molecular modeling and propose machine learning based strategies to boost simulations with ab initio accuracy to much larger scales than conventional approaches. We consider two representative challenges: 1) how to go from many-electron-ion to atomistic systems, for which the key has been a general and efficient representation of the potential energy surface generated by electronic structure models; 2) how to go from atomistic to coarse-grained systems, for which one is interested in the free energy of the coarse-grained variables as well as the associated dynamical behavior. Our strategies follow two seemingly obvious but non-trivial principles: 1) machine learning based models should respect important physical constraints like symmetry; 2) to build truly reliable models, efficient algorithms are needed to construct a minimal but truly representative training data set. We use these principles to construct the Deep Potential model for the potential energy surface, the Deep Potential Molecular dynamics (DeePMD) which is a new paradigm for performing ab initio molecular dynamics, a concurrent learning scheme (DP-GEN) for generating the data set on the fly, algorithms for constructing the Wannier centers (Deep Wanner) and for efficiently exploring the free energy landscape (Reinforced Dynamics), as well as a machine learning-based coarse grained molecular dynamics model (DeePCG), etc.Applications of these models and algorithms are presented for problems in chemistry, biology, and materials science. Finally, we present our efforts on developing related open-source software packages, which have now been widely used worldwide by experts and practitioners in the molecular simulation community.}, - author = {Zhang, L}, - date = {2020}, - title = {Machine Learning for Multi-Scale Molecular Modeling: Theories, Algorithms, and Applications}, - url = {https://search.proquest.com/openview/58ad7a1fdcc88005de25b81cd4cdc5d8/1?pq-origsite=gscholar&cbl=51922&diss=y} -} - -@article{zhangAcceleratingAtomisticSimulations2021a, - abstract = {Recently, machine learning methods have become easy-to-use tools for constructing high-dimensional interatomic potentials with ab initio accuracy. Although machine-learned interatomic potentials are generally orders of magnitude faster than first-principles calculations, they remain much slower than classical force fields, at the price of using more complex structural descriptors. To bridge this efficiency gap, we propose an embedded atom neural network approach with simple piecewise switching function-based descriptors, resulting in a favorable linear scaling with the number of neighbor atoms. Numerical examples validate that this piecewise machine-learning model can be over an order of magnitude faster than various popular machine-learned potentials with comparable accuracy for both metallic and covalent materials, approaching the speed of the fastest embedded atom method (i.e. several ms per atom per CPU core). The extreme efficiency of this approach promises its potential in first-principles atomistic simulations of very large systems and/or in a long timescale.}, - annotation = {WOS:000612961700004}, - author = {Zhang, Yaolong and Hu, Ce and Jiang, Bin}, - date = {2021-01-21}, - doi = {10.1039/d0cp05089j}, - issn = {1463-9076}, - journaltitle = {Physical Chemistry Chemical Physics}, - langid = {english}, - location = {{Cambridge}}, - number = {3}, - pages = {1815--1821}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Phys. Chem. Chem. Phys.}, - title = {Accelerating Atomistic Simulations with Piecewise Machine-Learned Ab {{Initio}} Potentials at a Classical Force Field-like Cost}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {23} -} - -@article{zhangActiveLearningUniformly2019, - abstract = {An active learning procedure called deep potential generator (DP-GEN) is proposed for the construction of accurate and transferable machine learning-based models of the potential energy surface (PES) for the molecular modeling of materials. This procedure consists of three main components: exploration, generation of accurate reference data, and training. Application to the sample systems of Al, Mg, and Al-Mg alloys demonstrates that DP-GEN can produce uniformly accurate PES models with a minimal number of reference data.}, - annotation = {WOS:000459927600002}, - author = {Zhang, Linfeng and Lin, De-Ye and Wang, Han and Car, Roberto and E, Weinan}, - date = {2019-02-25}, - doi = {10.1103/PhysRevMaterials.3.023804}, - issn = {2475-9953}, - journaltitle = {Physical Review Materials}, - langid = {english}, - location = {{College Pk}}, - number = {2}, - pages = {023804}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Mater.}, - title = {Active Learning of Uniformly Accurate Interatomic Potentials for Materials Simulation}, - url = {https://www.webofscience.com/wos/alldb/summary/ffcce553-2513-49c3-88af-d36432b825f2-03389e3a/times-cited-descending/1}, - urldate = {2021-08-05}, - volume = {3} -} - -@article{zhangAdaptiveCouplingDeep2018, - abstract = {An adaptive modeling method (AMM) that couples a deep neural network potential and a classical force field is introduced to address the accuracy-efficiency dilemma faced by the molecular simulation community. The AMM simulated system is decomposed into three types of regions. The first type captures the important phenomena in the system and requires high accuracy, for which we use the Deep Potential Molecular Dynamics (DeePMD) model in this work. The DeePMD model is trained to accurately reproduce the statistical properties of the ab initio molecular dynamics. The second type does not require high accuracy, and a classical force field is used to describe it in an efficient way. The third type is used for a smooth transition between the first and the second types of regions. By using a force interpolation scheme and imposing a thermodynamics force in the transition region, we make the DeePMD region embedded in the AMM simulated system as if it were embedded in a system that is fully described by the accurate potential. A representative example of the liquid water system is used to show the feasibility and promise of this method.}, - author = {Zhang, Linfeng and Wang, Han and E, Weinan}, - date = {2018}, - doi = {10.1063/1.5042714}, - journaltitle = {The Journal of chemical physics}, - number = {15}, - pages = {154107}, - publisher = {{AIP Publishing LLC}}, - title = {Adaptive Coupling of a Deep Neural Network Potential to a Classical Force Field}, - url = {https://aip.scitation.org/doi/abs/10.1063/1.5042714}, - volume = {149} -} - -@article{zhangAnomalousPhaseSeparation2021, - abstract = {We show that the celebrated Falicov-Kimball model exhibits rich and intriguing phase-ordering dynamics. Applying modern machine learning methods to enable large-scale quantum kinetic Monte Carlo simulations, we uncover an unusual phase-separation scenario in which the growth of charge checkerboard clusters competes with domain coarsening related to a hidden symmetry-breaking. A self-trapping mechanism as a result of this competition gives rise to arrested growth of checkerboard patterns and their super-clusters. Glassy behaviors similar to the one reported in this work could be generic for other correlated electron systems.}, - archiveprefix = {arXiv}, - author = {Zhang, Sheng and Zhang, Puhan and Chern, Gia-Wei}, - date = {2021-05-27}, - eprint = {2105.13304}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 6 pages, 5 figures}, - primaryclass = {cond-mat}, - title = {Anomalous Phase Separation and Hidden Coarsening of Super-Clusters in the {{Falicov}}-{{Kimball}} Model}, - url = {http://arxiv.org/abs/2105.13304}, - urldate = {2021-08-11} -} - -@article{zhangArrestedPhaseSeparation2021a, - abstract = {We present large-scale dynamical simulations of electronic phase separation in the single-band double-exchange model based on deep-learning neural-network potentials trained from small-size exact diagonalization solutions. We uncover an intriguing correlation-induced freezing behavior as doped holes are segregated from half-filled insulating background during equilibration. While the aggregation of holes is stabilized by the formation of ferromagnetic clusters through Hund's coupling between charge carriers and local magnetic moments, this stabilization also creates confining potentials for holes when antiferromagnetic spin-spin correlation is well developed in the background. The dramatically reduced mobility of the self-trapped holes prematurely disrupts further growth of the ferromagnetic clusters, leading to an arrested phase separation. Implications of our findings for phase separation dynamics in materials that exhibit colossal magnetoresistance effect are discussed.}, - archiveprefix = {arXiv}, - author = {Zhang, Puhan and Chern, Gia-Wei}, - date = {2021-05-17}, - eprint = {2105.08221}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {cond-mat}, - shorttitle = {Arrested Phase Separation in Double-Exchange Models}, - title = {Arrested Phase Separation in Double-Exchange Models: Machine-Learning Enabled Large-Scale Simulation}, - url = {http://arxiv.org/abs/2105.08221}, - urldate = {2021-08-10} -} - -@article{zhangBridgingGapDirect2019a, - abstract = {Direct dynamics simulations become increasingly popular in studying reaction dynamics for complex systems where analytical potential energy surfaces (PESs) are unavailable. Yet, the number and/or the propagation time of trajectories are often limited by high computational costs, and numerous energies and forces generated on-the-fly become wasted after simulations. We demonstrate here an example of reusing only a very small portion of existing direct dynamics data to reconstruct a 90-dimensional globally accurate reactive PES describing the interaction of CO2 with a movable Ni(100) surface based on a machine learning approach. In addition to reproducing previous results with much better statistics, we predict scattering probabilities of CO2 at the state-tostate level, which is extremely demanding for direct dynamics. We propose this unified way to investigate gaseous and gas-surface reactions of medium size, initiating with hundreds of preliminary direct dynamics trajectories, followed by low-cost and high-quality simulations on full-dimensional analytical PESs.}, - annotation = {WOS:000463678800005}, - author = {Zhang, Yaolong and Zhou, Xueyao and Jiang, Bin}, - date = {2019-03-21}, - doi = {10.1021/acs.jpclett.9b00085}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {6}, - pages = {1185--1191}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - title = {Bridging the {{Gap}} between {{Direct Dynamics}} and {{Globally Accurate Reactive Potential Energy Surfaces Using Neural Networks}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {10} -} - -@article{zhangCrystallizationP3Sn4Phase2021a, - abstract = {We performed molecular dynamics simulations to study the crystallization of the P3Sn4 phase from P(2)Sn(5 )liquid using a machine learning (ML) interatomic potential with desirable efficiency and accuracy. Our results capture the liquid properties of P2Sn5 at 1300 K, which is well above the melting temperature. The phase separation and crystallization are observed when P2Sn5 liquid is cooled down below 832 and 505 K, respectively. The simulation results are in good agreement with the experimentally observed phase transformation behaviors and provide useful insights into the complex nucleation and crystallization process at the details of atomistic scale. Our work also demonstrated that ML interatomic potentials based on neural network deep learning are robust and capable of accurately describing the energetics and kinetics of complex materials through molecular dynamics simulations.}, - annotation = {WOS:000619760700033}, - author = {Zhang, Chao and Sun, Yang and Wang, Hai-Di and Zhang, Feng and Wen, Tong-Qi and Ho, Kai-Ming and Wang, Cai-Zhuang}, - date = {2021-02-11}, - doi = {10.1021/acs.jpcc.0c08873}, - issn = {1932-7447}, - journaltitle = {Journal of Physical Chemistry C}, - langid = {english}, - location = {{Washington}}, - number = {5}, - pages = {3127--3133}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. C}, - title = {Crystallization of the {{P3Sn4 Phase}} upon {{Cooling P2Sn5 Liquid}} by {{Molecular Dynamics Simulation Using}} a {{Machine Learning Interatomic Potential}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {125} -} - -@article{zhangDeePCGConstructingCoarsegrained2018, - abstract = {We introduce a general framework for constructing coarse-grained potential models without ad hoc approximations such as limiting the potential to two- and/or three-body contributions. The scheme, called the Deep Coarse-Grained Potential (abbreviated DeePCG), exploits a carefully crafted neural network to construct a many-body coarse-grained potential. The network is trained with full atomistic data in a way that preserves the natural symmetries of the system. The resulting model is very accurate and can be used to sample the configurations of the coarse-grained variables in a much faster way than with the original atomistic model. As an application, we consider liquid water and use the oxygen coordinates as the coarse-grained variables, starting from a full atomistic simulation of this system at the ab initio molecular dynamics level. We find that the two-body, three-body, and higher-order oxygen correlation functions produced by the coarse-grained and full atomistic models agree very well with each other, illustrating the effectiveness of the DeePCG model on a rather challenging task.}, - archiveprefix = {arXiv}, - arxivid = {1802.08549v3}, - author = {Zhang, Linfeng and Han, Jiequn and Wang, Han and Car, Roberto and Weinan, Weinan E.}, - date = {2018-07}, - doi = {10.1063/1.5027645}, - eprint = {1802.08549v3}, - eprinttype = {arxiv}, - number = {3}, - publisher = {{American Institute of Physics Inc.}}, - title = {{{DeePCG}}: {{Constructing}} Coarse-Grained Models via Deep Neural Networks}, - url = {https://aip.scitation.org/doi/abs/10.1063/1.5027645}, - volume = {149} -} - -@article{zhangDeepNeuralNetwork2020, - abstract = {We introduce a deep neural network to model in a symmetry preserving way the environmental dependence of the centers of the electronic charge. The model learns from ab initio density functional theory, wherein the electronic centers are uniquely assigned by the maximally localized Wannier functions. When combined with the deep potential model of the atomic potential energy surface, the scheme predicts the dielectric response of insulators for trajectories inaccessible to direct ab initio simulation. The scheme is nonperturbative and can capture the response of a mutating chemical environment. We demonstrate the approach by calculating the infrared spectra of liquid water at standard conditions, and of ice under extreme pressure, when it transforms from a molecular to an ionic crystal.}, - annotation = {WOS:000550993300002}, - author = {Zhang, Linfeng and Chen, Mohan and Wu, Xifan and Wang, Han and Weinan, E. and Car, Roberto}, - date = {2020-07-22}, - doi = {10.1103/PhysRevB.102.041121}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {4}, - pages = {041121}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Deep Neural Network for the Dielectric Response of Insulators}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {102} -} - -@article{zhangDeepPotentialMolecular2018, - abstract = {We introduce a scheme for molecular simulations, the deep potential molecular dynamics (DPMD) method, based on a many-body potential and interatomic forces generated by a carefully crafted deep neural network trained with ab initio data. The neural network model preserves all the natural symmetries in the problem. It is first-principles based in the sense that there are no ad hoc components aside from the network model. We show that the proposed scheme provides an efficient and accurate protocol in a variety of systems, including bulk materials and molecules. In all these cases, DPMD gives results that are essentially indistinguishable from the original data, at a cost that scales linearly with system size.}, - annotation = {WOS:000429119100003}, - author = {Zhang, Linfeng and Han, Jiequn and Wang, Han and Car, Roberto and Weinan, E.}, - date = {2018-04-04}, - doi = {10.1103/PhysRevLett.120.143001}, - issn = {0031-9007}, - journaltitle = {Physical Review Letters}, - langid = {english}, - location = {{College Pk}}, - number = {14}, - pages = {143001}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Lett.}, - shorttitle = {Deep {{Potential Molecular Dynamics}}}, - title = {Deep {{Potential Molecular Dynamics}}: {{A Scalable Model}} with the {{Accuracy}} of {{Quantum Mechanics}}}, - url = {https://www.webofscience.com/wos/alldb/summary/ffcce553-2513-49c3-88af-d36432b825f2-03389e3a/times-cited-descending/1}, - urldate = {2021-08-05}, - volume = {120} -} - -@article{zhangDPGENConcurrentLearning2020, - abstract = {In recent years, promising deep learning based interatomic potential energy surface (PES) models have been proposed that can potentially allow us to perform molecular dynamics simulations for large scale systems with quantum accuracy. However, making these models truly reliable and practically useful is still a very non-trivial task. A key component in this task is the generation of datasets used in model training. In this paper, we introduce the Deep Potential GENerator (DP-GEN), an open-source software platform that implements the recently proposed "on-the-fly" learning procedure (Zhang et al. 2019) and is capable of generating uniformly accurate deep learning based PES models in a way that minimizes human intervention and the computational cost for data generation and model training. DP-GEN automatically and iteratively performs three steps: exploration, labeling, and training. It supports various popular packages for these three steps: LAMMPS for exploration, Quantum Espresso, VASP, CP2K, etc. for labeling, and DeePMD-kit for training. It also allows automatic job submission and result collection on different types of machines, such as high performance clusters and cloud machines, and is adaptive to different job management tools, including Slurm, PBS, and LSF. As a concrete example, we illustrate the details of the process for generating a general-purpose PES model for Cu using DP-GEN. Program summary Program Title: DP-GEN Program Files doi: http://dx.dot.org/10.17632/sxybkgc5xc.1 Licensing provisions: LGPL Programming language: Python Nature of problem: Generating reliable deep learning based potential energy models with minimal human intervention and computational cost. Solution method: The concurrent learning scheme is implemented. Supports for sampling configuration space with LAMMPS, generating ab initio data with Quantum Espresso, VASP, CP2K and training potential models with DeePMD-kit are provided. Supports for different machines including workstations, high performance clusters and cloud machines are provided. Supports for job management tools including Slurm, PBS, LSF are provided. (C) 2020 Elsevier B.V. All rights reserved.}, - annotation = {WOS:000537843600017}, - author = {Zhang, Yuzhi and Wang, Haidi and Chen, Weijie and Zeng, Jinzhe and Zhang, Linfeng and Wang, Han and Weinan, E.}, - date = {2020-08}, - doi = {10.1016/j.cpc.2020.107206}, - issn = {0010-4655}, - journaltitle = {Computer Physics Communications}, - langid = {english}, - location = {{Amsterdam}}, - pages = {107206}, - publisher = {{Elsevier}}, - shortjournal = {Comput. Phys. Commun.}, - shorttitle = {{{DP}}-{{GEN}}}, - title = {{{DP}}-{{GEN}}: {{A}} Concurrent Learning Platform for the Generation of Reliable Deep Learning Based Potential Energy Models}, - url = {https://www.sciencedirect.com/science/article/abs/pii/S001046552030045X}, - urldate = {2021-08-06}, - volume = {253} -} - -@article{zhangEfficientAccurateSimulations2020, - abstract = {Machine learning has revolutionized the high-dimensional representations for molecular properties such as potential energy. However, there are scarce machine learning models targeting tensorial properties, which are rotationally covariant. Here, we propose tensorial neural network (NN) models to learn both tensorial response and transition properties, in which atomic coordinate vectors are multiplied with scalar NN outputs or their derivatives to preserve the rotationally covariant symmetry. This strategy keeps structural descriptors symmetry invariant so that the resulting tensorial NN models are as efficient as their scalar counterparts. We validate the performance and universality of this approach by learning response properties of water oligomers and liquid water, and transition dipole moment of a model structural unit of proteins. Machine learned tensorial models have enabled efficient simulations of vibrational spectra of liquid water and ultraviolet spectra of realistic proteins, promising feasible and accurate spectroscopic simulations for biomolecules and materials.}, - author = {Zhang, Yaolong and Ye, Sheng and Zhang, Jinxiao and Hu, Ce and Jiang, Jun and Jiang, Bin}, - date = {2020}, - doi = {10.1021/acs.jpcb.0c06926}, - journaltitle = {The Journal of Physical Chemistry B}, - number = {33}, - pages = {7284--7290}, - publisher = {{ACS Publications}}, - title = {Efficient and Accurate Simulations of Vibrational and Electronic Spectra with Symmetry-Preserving Neural Network Models for Tensorial Properties}, - volume = {124} -} - -@article{zhangEmbeddedAtomNeural2019, - abstract = {We propose a simple, but efficient and accurate, machine learning (ML) model for developing a high-dimensional potential energy surface. This so-called embedded atom neural network (EANN) approach is inspired by the well-known empirical embedded atom method (EAM) model used in the condensed phase. It simply replaces the scalar embedded atom density in EAM with a Gaussian-type orbital based density vector and represents the complex relationship between the embedded density vector and atomic energy by neural networks. We demonstrate that the EANN approach is equally accurate as several established ML models in representing both big molecular and extended periodic systems, yet with much fewer parameters and configurations. It is highly efficient as it implicitly contains the three-body information without an explicit sum of the conventional costly angular descriptors. With high accuracy and efficiency, EANN potentials can vastly accelerate molecular dynamics and spectroscopic simulations in complex systems at ab initio level.}, - annotation = {WOS:000484884300030}, - author = {Zhang, Yaolong and Hu, Ce and Jiang, Bin}, - date = {2019-09-05}, - doi = {10.1021/acs.jpclett.9b02037}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {17}, - pages = {4962--4967}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - shorttitle = {Embedded {{Atom Neural Network Potentials}}}, - title = {Embedded {{Atom Neural Network Potentials}}: {{Efficient}} and {{Accurate Machine Learning}} with a {{Physically Inspired Representation}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {10} -} - -@article{zhangEmbeddedAtomNeural2019, - abstract = {We propose a simple, but efficient and accurate, machine learning (ML) model for developing a high-dimensional potential energy surface. This so-called embedded atom neural network (EANN) approach is inspired by the well-known empirical embedded atom method (EAM) model used in the condensed phase. It simply replaces the scalar embedded atom density in EAM with a Gaussian-type orbital based density vector and represents the complex relationship between the embedded density vector and atomic energy by neural networks. We demonstrate that the EANN approach is equally accurate as several established ML models in representing both big molecular and extended periodic systems, yet with much fewer parameters and configurations. It is highly efficient as it implicitly contains the three-body information without an explicit sum of the conventional costly angular descriptors. With high accuracy and efficiency, EANN potentials can vastly accelerate molecular dynamics and spectroscopic simulations in complex systems at ab initio level.}, - annotation = {WOS:000484884300030}, - author = {Zhang, Yaolong and Hu, Ce and Jiang, Bin}, - date = {2019-09-05}, - doi = {10.1021/acs.jpclett.9b02037}, - issn = {1948-7185}, - journaltitle = {Journal of Physical Chemistry Letters}, - langid = {english}, - location = {{Washington}}, - number = {17}, - pages = {4962--4967}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. Lett.}, - shorttitle = {Embedded {{Atom Neural Network Potentials}}}, - title = {Embedded {{Atom Neural Network Potentials}}: {{Efficient}} and {{Accurate Machine Learning}} with a {{Physically Inspired Representation}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {10} -} - -@misc{zhangEndtoendSymmetryPreserving2018, - abstract = {Machine learning models are changing the paradigm of molecular modeling, which is a fundamental tool for material science, chemistry, and computational biology. Of particular interest is the inter-atomic potential energy surface (PES). Here we develop Deep Potential - Smooth Edition (DeepPot-SE), an end-to-end machine learning-based PES model, which is able to efficiently represent the PES for a wide variety of systems with the accuracy of ab initio quantum mechanics models. By construction, DeepPot-SE is extensive and continuously differentiable, scales linearly with system size, and preserves all the natural symmetries of the system. Further, we show that DeepPot-SE describes finite and extended systems including organic molecules, metals, semiconductors, and insulators with high fidelity.}, - author = {Zhang, Linfeng and Han, Jiequn and {Wang, Han} and {Wissam A. Saidi} and {Roberto Car} and E, Weinan}, - date = {2018}, - doi = {arXiv:1805.09003}, - title = {End-to-End Symmetry Preserving Inter-Atomic Potential Energy Model for Finite and Extended Systems}, - url = {https://arxiv.org/abs/1805.09003}, - urldate = {2021-08-09} -} - -@article{zhangGlobalOptimizationChemical2021, - abstract = {Chemical clusters are relevant to many applications in catalysis, separations, materials, and energy sciences. Experimentally, the structure of clusters is difficult to determine, but it is very important in understanding their chemistry and properties. Computational methods can be used to examine cluster structure, however finding the most stable structure is not simple, particularly as the cluster size increases. Global optimization techniques have long been used to tackle the problem of the most stable structure, but such approaches would have to look for a global minimum, while sampling local minima over the whole potential energy surface as well. In this review, the state-of-the-art theory of global optimization theory is summarized. First, the definition, significance, relation to experiments, and a brief history of global optimization is presented. We then discuss, in more detail, three versatile global optimization methods: the basin hopping, the artificial bee colony algorithm, and the genetic algorithm. We close with some representative application examples of global optimization of clusters since 2016 and the challenges, open questions and opportunities in this field.}, - annotation = {WOS:000591191600001}, - author = {Zhang, Jun and Glezakou, Vassiliki-Alexandra}, - date = {2021-04-05}, - doi = {10.1002/qua.26553}, - issn = {0020-7608}, - journaltitle = {International Journal of Quantum Chemistry}, - langid = {english}, - location = {{Hoboken}}, - number = {7}, - pages = {e26553}, - publisher = {{Wiley}}, - shortjournal = {Int. J. Quantum Chem.}, - shorttitle = {Global Optimization of Chemical Cluster Structures}, - title = {Global Optimization of Chemical Cluster Structures: {{Methods}}, Applications, and Challenges}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {121} -} - -@article{zhangIsotopeEffectsXray2020, - abstract = {The isotope effects in x-ray absorption spectra of liquid water are studied by a many-body approach within electron-hole excitation theory. The molecular structures of both light and heavy water are modeled by path-integral molecular dynamics based on the advanced deep-learning technique. The neural network is trained on ab initio data obtained with SCAN density functional theory. The experimentally observed isotope effect in x-ray absorption spectra is reproduced semiquantitatively in theory. Compared to the spectrum in normal water, the blueshifted and less pronounced pre- and main-edge in heavy water reflect that the heavy water is more structured at short- and intermediate-range of the hydrogen-bond network. In contrast, the isotope effect on the spectrum is negligible at post-edge, which is consistent with the identical long-range ordering in both liquids as observed in the diffraction experiment.}, - annotation = {WOS:000572817400002}, - author = {Zhang, Chunyi and Zhang, Linfeng and Xu, Jianhang and Tang, Fujie and Santra, Biswajit and Wu, Xifan}, - date = {2020-09-25}, - doi = {10.1103/PhysRevB.102.115155}, - issn = {2469-9950}, - journaltitle = {Physical Review B}, - langid = {english}, - location = {{College Pk}}, - number = {11}, - pages = {115155}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. B}, - title = {Isotope Effects in X-Ray Absorption Spectra of Liquid Water}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {102} -} - -@article{zhangLinearFrequencyPrinciple2021a, - abstract = {Why heavily parameterized neural networks (NNs) do not overfit the data is an important long standing open question. We propose a phenomenological model of the NN training to explain this non-overfitting puzzle. Our linear frequency principle (LFP) model accounts for a key dynamical feature of NNs: they learn low frequencies first, irrespective of microscopic details. Theory based on our LFP model shows that low frequency dominance of target functions is the key condition for the non-overfitting of NNs and is verified by experiments. Furthermore, through an ideal two-layer NN, we unravel how detailed microscopic NN training dynamics statistically gives rise to an LFP model with quantitative prediction power.}, - annotation = {WOS:000632901300001}, - author = {Zhang, Yaoyu and Luo, Tao and Ma, Zheng and Xu, Zhi-Qin John}, - date = {2021-03}, - doi = {10.1088/0256-307X/38/3/038701}, - issn = {0256-307X}, - journaltitle = {Chinese Physics Letters}, - langid = {english}, - location = {{Bristol}}, - number = {3}, - pages = {038701}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Chin. Phys. Lett.}, - title = {A {{Linear Frequency Principle Model}} to {{Understand}} the {{Absence}} of {{Overfitting}} in {{Neural Networks}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {38} -} - -@article{zhangMachineLearningDynamics2020, - abstract = {We demonstrate machine-learning enabled large-scale dynamical simulations of electronic phase separation in double-exchange system. This model, also known as the ferromagnetic Kondo lattice model, is believed to be relevant for the colossal magnetoresistance phenomenon. Real-space simulations of such inhomogeneous states with exchange forces computed from the electron Hamiltonian can be prohibitively expensive for large systems. Here we show that linear-scaling exchange field computation can be achieved using neural networks trained by datasets from exact calculation on small lattices. Our Landau-Lifshitz dynamics simulations based on machine-learning potentials nicely reproduce not only the nonequilibrium relaxation process, but also correlation functions that agree quantitatively with exact simulations. Our work paves the way for large-scale dynamical simulations of correlated electron systems using machine-learning models.}, - archiveprefix = {arXiv}, - author = {Zhang, Puhan and Saha, Preetha and Chern, Gia-Wei}, - date = {2020-06-07}, - eprint = {2006.04205}, - eprinttype = {arxiv}, - langid = {english}, - note = {Comment: 6 pages, 4 figures}, - primaryclass = {cond-mat}, - title = {Machine Learning Dynamics of Phase Separation in Correlated Electron Magnets}, - url = {http://arxiv.org/abs/2006.04205}, - urldate = {2021-08-11} -} - -@article{zhangMolecularCTUnifying, - abstract = {Deep learning is changing many areas in molecular physics, and it has shown great potential to deliver new solutions to challenging molecular modeling problems. Along with this trend arises the increasing demand of expressive and versatile neural network architectures which are compatible with molecular systems. A new deep neural network architecture, Molecular Configuration Transformer (Molecular CT), is introduced for this purpose. Molecular CT is composed of a relation-aware encoder module and a computationally universal geometry learning unit, thus able to account for the relational constraints between particles meanwhile scalable to different particle numbers and invariant w.r.t. the trans-rotational transforms. The computational efficiency and universality make Molecular CT versatile for a variety of molecular learning scenarios and especially appealing for transferable representation learning across different molecular systems. As examples, we show that Molecular CT enables representational learning for molecular systems at different scales, and achieves comparable or improved results on common benchmarks using a more lightweighted structure compared to baseline models.}, - author = {Zhang, Jun and Zhou, Yaqiang and Lei, Yao-Kun and Yang, Yi Isaac and Gao, Yi Qin}, - langid = {english}, - pages = {14}, - title = {Molecular {{CT}}: {{Unifying Geometry}} and {{Representation Learning}} for {{Molecules}} at {{Different Scales}}} -} - -@article{zhangMongeampBackslashEre2018, - abstract = {We present a deep generative model, named Monge-Ampère flow, which builds on continuous-time gradient flow arising from the Monge-Ampère equation in optimal transport theory. The generative map from the latent space to the data space follows a dynamical system, where a learnable potential function guides a compressible fluid to flow towards the target density distribution. Training of the model amounts to solving an optimal control problem. The Monge-Ampère flow has tractable likelihoods and supports efficient sampling and inference. One can easily impose symmetry constraints in the generative model by designing suitable scalar potential functions. We apply the approach to unsupervised density estimation of the MNIST dataset and variational calculation of the two-dimensional Ising model at the critical point. This approach brings insights and techniques from Monge-Ampère equation, optimal transport, and fluid dynamics into reversible flow-based generative models.}, - archiveprefix = {arXiv}, - author = {Zhang, Linfeng and Wang, Lei}, - date = {2018}, - eprint = {1809.10188}, - eprinttype = {arxiv}, - title = {Monge-Amp\$\textbackslash backslash\$ere Flow for Generative Modeling}, - url = {https://arxiv.org/abs/1809.10188} -} - -@article{zhangPerspectiveDeepLearning2020a, - abstract = {Deep learning is transforming many areas in science, and it has great potential in modeling molecular systems. However, unlike the mature deployment of deep learning in computer vision and natural language processing, its development in molecular modeling and simulations is still at an early stage, largely because the inductive biases of molecules are completely different from those of images or texts. Footed on these differences, we first reviewed the limitations of traditional deep learning models from the perspective of molecular physics and wrapped up some relevant technical advancement at the interface between molecular modeling and deep learning. We do not focus merely on the ever more complex neural network models; instead, we introduce various useful concepts and ideas brought by modern deep learning. We hope that transacting these ideas into molecular modeling will create new opportunities. For this purpose, we summarized several representative applications, ranging from supervised to unsupervised and reinforcement learning, and discussed their connections with the emerging trends in deep learning. Finally, we give an outlook for promising directions which may help address the existing issues in the current framework of deep molecular modeling.}, - annotation = {WOS:000566759400001}, - author = {Zhang, Jun and Lei, Yao-Kun and {hZang}, Zhen and Chang, Junhan and Li, Maodong and Han, Xu and Yang, Lijiang and Yang, Yi Isaac and Gao, Yi Qin}, - date = {2020-08-27}, - doi = {10.1021/acs.jpca.0c04473}, - issn = {1089-5639}, - journaltitle = {Journal of Physical Chemistry A}, - langid = {english}, - location = {{Washington}}, - number = {34}, - pages = {6745--6763}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. A}, - title = {A {{Perspective}} on {{Deep Learning}} for {{Molecular Modeling}} and {{Simulations}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {124} -} - -@article{zhangPhaseDiagramDeep2021, - abstract = {Using the Deep Potential methodology, we construct a model that reproduces accurately the potential energy surface of the SCAN approximation of density functional theory for water, from low temperature and pressure to about 2400 K and 50 GPa, excluding the vapor stability region. The computational efficiency of the model makes it possible to predict its phase diagram using molecular dynamics. Satisfactory overall agreement with experimental results is obtained. The fluid phases, molecular and ionic, and all the stable ice polymorphs, ordered and disordered, are predicted correctly, with the exception of ice III and XV that are stable in experiments, but metastable in the model. The evolution of the atomic dynamics upon heating, as ice VII transforms first into ice VII'' and then into an ionic fluid, reveals that molecular dissociation and breaking of the ice rules coexist with strong covalent fluctuations, explaining why only partial ionization was inferred in experiments.}, - annotation = {WOS:000661898600005}, - author = {Zhang, Linfeng and Wang, Han and Car, Roberto and Weinan, E.}, - date = {2021-06-09}, - doi = {10.1103/PhysRevLett.126.236001}, - issn = {0031-9007}, - journaltitle = {Physical Review Letters}, - langid = {english}, - location = {{College Pk}}, - number = {23}, - pages = {236001}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Lett.}, - title = {Phase {{Diagram}} of a {{Deep Potential Water Model}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {126} -} - -@article{zhangReinforcedDynamicsEnhanced2018, - abstract = {A new approach for efficiently exploring the configuration space and computing the free energy of large atomic and molecular systems is proposed, motivated by an analogy with reinforcement learning. There are two major components in this new approach. Like metadynamics, it allows for an efficient exploration of the configuration space by adding an adaptively computed biasing potential to the original dynamics. Like deep reinforcement learning, this biasing potential is trained on the fly using deep neural networks, with data collected judiciously from the exploration and an uncertainty indicator from the neural network model playing the role of the reward function. Parameterization using neural networks makes it feasible to handle cases with a large set of collective variables. This has the potential advantage that selecting precisely the right set of collective variables has now become less critical for capturing the structural transformations of the system. The method is illustrated by studying the full-atom, explicit solvent models of alanine dipeptide and tripeptide, as well as the system of a polyalanine-10 molecule with 20 collective variables.}, - author = {Zhang, Linfeng and Wang, Han and E, Weinan}, - date = {2018}, - doi = {10.1063/1.5019675}, - journaltitle = {The Journal of chemical physics}, - number = {12}, - pages = {124113}, - publisher = {{AIP Publishing LLC}}, - title = {Reinforced Dynamics for Enhanced Sampling in Large Atomic and Molecular Systems}, - url = {https://aip.scitation.org/doi/abs/10.1063/1.5019675}, - volume = {148} -} - -@article{zhangReinforcementLearningMultiScale, - abstract = {Molecular simulations are widely applied in the study of chemical and bio-physical systems of interest. However, the accessible timescales of atomistic simulations are limited, and extracting equilibrium properties of systems containing rare events remains challenging. Two distinct strategies are usually adopted in this regard: either sticking to the atomistic level and performing enhanced sampling, or trading details for speed by leveraging coarse-grained models. Although both strategies are promising, either of them, if adopted individually, exhibits severe limitations. In this paper we propose a machine-learning approach to take advantage of both strategies. In this approach, simulations on different scales are executed simultaneously and benefit mutually from their cross-talks: Accurate coarse-grained (CG) models can be inferred from the fine-grained (FG) simulations; In turn, FG simulations can be boosted by the guidance of CG models. Our method grounds on unsupervised and reinforcement learning, defined by a variational and adaptive training objective, and allows end-to-end training of parametric models. Through multiple experiments, we show that our method is efficient and flexible, and performs well on challenging chemical and bio-molecular systems.}, - author = {Zhang, Jun and Lei, Yao-Kun and Yang, Yi Isaac and Gao, Yi Qin}, - langid = {english}, - pages = {26}, - title = {Reinforcement {{Learning}} for {{Multi}}-{{Scale Molecular Modeling}}} -} - -@article{zhangTypeGeneralizationError, - abstract = {How initialization and loss function affect the learning of a deep neural network (DNN), specifically its generalization error, is an important problem in practice. In this work, by exploiting the linearity of DNN training dynamics in the NTK regime (Jacot et al., 2018; Lee et al., 2019), we provide an explicit and quantitative answer to this problem. Focusing on regression problem, we prove that, in the NTK regime, for any loss in a general class of functions, the DNN finds the same global minima—the one that is nearest to the initial value in the parameter space, or equivalently, the one that is closest to the initial DNN output in the corresponding reproducing kernel Hilbert space. Using these optimization problems, we quantify the impact of initial output and prove that a random non-zero one increases the generalization error. We further propose an antisymmetrical initialization (ASI) trick that eliminates this type of error and accelerates the training. To understand whether the above results hold in general, we also perform experiments for DNNs in the non-NTK regime, which demonstrate the effectiveness of our theoretical results and the ASI trick in a qualitative sense. Overall, our work serves as a baseline for the further investigation of the impact of initialization and loss function on the generalization of DNNs, which can potentially guide and improve the training of DNNs in practice.}, - author = {Zhang, Yaoyu and Xu, Zhi-Qin John and Luo, Tao and Ma, Zheng}, - langid = {english}, - pages = {21}, - title = {A Type of Generalization Error Induced by Initialization in Deep Neural Networks} -} - -@article{zhangWarmDenseMatter2020, - abstract = {Simulating warm dense matter that undergoes a wide range of temperatures and densities is challenging. Predictive theoretical models, such as quantum-mechanics-based first-principles molecular dynamics (FPMD), require a huge amount of computational resources. Herein, we propose a deep learning based scheme called electron temperature dependent deep potential molecular dynamics (TDDPMD), which can be readily applied to study larger systems with longer trajectories, yielding more accurate properties. We take warm dense beryllium (Be) as an example with the training data from FPMD simulations spanning a wide range of temperatures (0.4–2500\,eV) and densities (3.50–8.25\,g/cm3). The TDDPMD method well reproduces the principal Hugoniot curve and radial distribution functions from the FPMD method. Furthermore, it depicts the reflection point of the Hugoniot curve more smoothly and provides more converged diffusion coefficients. We also show the new model can yield static structure factors and dynamic structure factors of warm dense Be.}, - author = {Zhang, Yuzhi and Gao, Chang and Liu, Qianrui and Zhang, Linfeng and Wang, Han and Chen, Mohan}, - date = {2020}, - doi = {10.1063/5.0023265}, - journaltitle = {Physics of Plasmas}, - number = {12}, - pages = {122704}, - publisher = {{AIP Publishing LLC}}, - title = {Warm Dense Matter Simulation via Electron Temperature Dependent Deep Potential Molecular Dynamics}, - url = {https://aip.scitation.org/doi/abs/10.1063/5.0023265}, - volume = {27} -} - -@article{zhaoLearningPhysicsPattern2020a, - abstract = {Using a framework of partial differential equation-constrained optimization, we demonstrate that multiple constitutive relations can be extracted simultaneously from a small set of images of pattern formation. Examples include state-dependent properties in phase-field models, such as the diffusivity, kinetic prefactor, free energy, and direct correlation function, given only the general form of the Cahn-Hilliard equation, Allen-Cahn equation, or dynamical density functional theory (phase-field crystal model). Constraints can be added based on physical arguments to accelerate convergence and avoid spurious results. Reconstruction of the free energy functional, which contains nonlinear dependence on the state variable and differential or convolutional operators, opens the possibility of learning nonequilibrium thermodynamics from only a few snapshots of the dynamics.}, - annotation = {WOS:000513568400001}, - author = {Zhao, Hongbo and Storey, Brian D. and Braatz, Richard D. and Bazant, Martin Z.}, - date = {2020-02-14}, - doi = {10.1103/PhysRevLett.124.060201}, - issn = {0031-9007}, - journaltitle = {Physical Review Letters}, - langid = {english}, - location = {{College Pk}}, - number = {6}, - pages = {060201}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Phys. Rev. Lett.}, - title = {Learning the {{Physics}} of {{Pattern Formation}} from {{Images}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {124} -} - -@article{zhaoTheoreticalPredictionRedox2021, - abstract = {Redox potential is a crucial electrochemical parameter in the electrorefining process of spent fuel reprocessing. Unfortunately, the harsh measurement environment of spent fuel makes it difficult to obtain. With the continuous updating of computing technology, people have developed the method for calculating the redox potentials using first-principles molecular dynamics (FPMD), but limited by the calculation cost, the simulation scale and time of FPMD are restricted. To make the calculation results more convincing, this work used deep potential (DP) to realize redox potential calculation on a larger time scale. We extracted datasets from FPMD calculations and used these to train and validate the DP, and compared energies, forces, and radial distribution functions that are evaluated using DFT and DP, to demonstrate that DP can achieve DFT accuracy. Using La3+/La as the reference electrode, the redox potentials of Ce3+/Ce, Pr3+/Pr, and Y3+/Y in the LiCl-KCl mixed molten salt system at 723 K were calculated. The results matched well with the FPMD results and experimental data. This work fully demonstrates the feasibility of the DP in calculating the redox potentials. Simultaneously, it provides new idea for obtaining accurate data in the process of spent fuel reprocessing.}, - author = {Zhao, Jia and Liang, Wenshuo and Lu, Guimin}, - date = {2021-05-01}, - doi = {10/gmfwvw}, - issn = {1862-0760}, - journaltitle = {Ionics}, - langid = {english}, - number = {5}, - pages = {2079--2088}, - shortjournal = {Ionics}, - title = {Theoretical Prediction on the Redox Potentials of Rare-Earth Ions by Deep Potentials}, - url = {https://doi.org/10.1007/s11581-021-03988-0}, - urldate = {2021-08-11}, - volume = {27} -} - -@article{zhengRetentionRecyclingDeuterium2021a, - abstract = {Understanding the retention and recycling of hydrogen isotopes in liquid metal plasma-facing materials such as liquid Li, Sn, and Li-Sn are of fundamental importance in designing magnetically confined fusion reactors. We perform first-principles molecules dynamics simulations of liquid Li-Sn slab with inserted D atoms to provide microscopic insights into the interactions of D with Li-Sn liquid metal. We prepare two samples with low and high concentrations of D atoms. We observe evaporation of D molecules out of the Li-Sn slabs in both concentrations of D. With detailed analysis, we unveil a cooperative process of forming D-2 molecules in liquid Li-Sn, where Li atoms act as catalytic centers to trap a D atom before another D comes nearby to form a molecule, and the surplus charges are transferred from D-2 to nearby Sn atoms. Furthermore, we predict a temperature window in the low concentration case in which D-2 molecules can escape to vacuum, while LiD molecules cannot. The above findings deepen our understanding of interactions between hydrogen isotopes and Li-Sn liquid metal. (C) 2020 Elsevier B.V. All rights reserved.}, - annotation = {WOS:000600309100006}, - author = {Zheng, Daye and Shen, Zhen-Xiong and Chen, Mohan and Ren, Xinguo and He, Lixin}, - date = {2021-01}, - doi = {10.1016/j.jnucmat.2020.152542}, - issn = {0022-3115}, - journaltitle = {Journal of Nuclear Materials}, - langid = {english}, - location = {{Amsterdam}}, - pages = {152542}, - publisher = {{Elsevier}}, - shortjournal = {J. Nucl. Mater.}, - title = {Retention and Recycling of Deuterium in Liquid Lithium-Tin Slab Studied by First-Principles Molecular Dynamics}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {543} -} - -@article{zhouAtomicstatedependentScreeningModel2021a, - abstract = {An ion embedded in warm/hot dense plasmas will greatly alter its microscopic structure and dynamics, as well as the macroscopic radiation transport properties of the plasmas, due to complicated many-body interactions with surrounding particles. Accurate theoretically modeling of such kind of quantum many-body interactions is essential but very challenging. In this work, we propose an atomic-state-dependent screening model for treating the plasmas with a wide range of temperatures and densities, in which the contributions of three-body recombination processes are included. We show that the electron distributions around an ion are strongly correlated with the ionic state studied due to the contributions of three-body recombination processes. The feasibility and validation of the proposed model are demonstrated by reproducing the experimental result of the line-shift of hot-dense plasmas as well as the classical molecular dynamic simulations of moderately coupled ultra-cold neutral plasmas. Our work opens a promising way to treat the screening effect of hot and warm dense plasma, which is a bottleneck of those extensive studies in high-energy-density physics, such as atomic processes in plasma, plasma spectra and radiation transport properties, among others. Atoms embedded in dense hot plasmas are affected by complex many-body interactions which challenges our capacity to model high energy density plasma. The authors propose a solution to the effects of many-body interactions on ions in dense plasmas, with a particular focus on the threebody interaction.}, - annotation = {WOS:000668539100001}, - author = {Zhou, Fuyang and Qu, Yizhi and Gao, Junwen and Ma, Yulong and Wu, Yong and Wang, Jianguo}, - date = {2021-06-30}, - doi = {10.1038/s42005-021-00652-x}, - issn = {2399-3650}, - journaltitle = {Communications Physics}, - langid = {english}, - location = {{Berlin}}, - number = {1}, - pages = {148}, - publisher = {{Nature Research}}, - shortjournal = {Commun. Phys.}, - title = {Atomic-State-Dependent Screening Model for Hot and Warm Dense Plasmas}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {4} -} - -@article{zhouFrameindependentVectorcloudNeural2021, - abstract = {Constitutive models are widely used for modeling complex systems in science and engineering, where first principle-based, well-resolved simulations are often prohibitively expensive. For example, in uid dynamics, constitutive models are required to describe nonlocal, unresolved physics such as turbulence and laminar turbulent transition. However, traditional constitutive models based on partial differential equations (PDEs) often lack robustness and are too rigid to accommodate diverse calibration datasets. We propose a frame independent, nonlocal constitutive model based on a vector-cloud neural network that can be learned with data. The model predicts the closure variable at a point based on the ow information in its neighborhood. Such nonlocal information is represented by a group of points, each having a feature vector attached to it, and thus the input is referred to as vector cloud. The cloud is mapped to the closure variable through a frame-independent neural network, invariant both to coordinate translation and rotation and to the ordering of points in the cloud. As such, the network can deal with any number of arbitrarily arranged grid points and thus is suitable for unstructured meshes in uid simulations. The merits of the proposed network are demonstrated for scalar transport PDEs on a family of parameterized periodic hill geometries. The vectorcloud neural network is a promising tool not only as nonlocal constitutive models and but also as general surrogate models for PDEs on irregular domains.}, - archiveprefix = {arXiv}, - author = {Zhou, Xu-Hui and Han, Jiequn and Xiao, Heng}, - date = {2021}, - eprint = {2103.06685}, - eprinttype = {arxiv}, - title = {Frame-Independent Vector-Cloud Neural Network for Nonlocal Constitutive Modelling on Arbitrary Grids}, - url = {https://arxiv.org/pdf/2103.06685} -} - -@article{zhouStructureDynamicsSupercooled2021, - abstract = {Studies of supercooled liquid phase-change materials are important for the development of phase-change memory and neuromorphic computing devices. Here, we use a machine-learning-based interatomic potential for Ge2Sb2Te5 (GST) to carry out large-scale molecular-dynamics simulations of liquid and supercooled liquid Ge2Sb2Te5. We demonstrate a pronounced effect of the thermostat parameters on the simulation results, and we show how using a Langevin thermostat with optimized damping values can lead to excellent agreement with reference ab initio molecular dynamics (AIMD) simulations. Structural and dynamical analyses are presented, including studies of radial and angular distributions, homopolar bonds, and the temperature-dependent diffusivity. Our work demonstrates the usefulness of ML-driven molecular dynamics for further studies of supercooled liquid GST, with length and time scales far exceeding those that would be accessible to AIMD.}, - author = {Zhou, Yu-Xing and Zhang, Han-Yi and Deringer, Volker L. and Zhang, Wei}, - date = {2021-03}, - doi = {10/gmf6g6}, - issn = {1862-6254, 1862-6270}, - journaltitle = {physica status solidi (RRL) – Rapid Research Letters}, - langid = {english}, - number = {3}, - pages = {2000403}, - shortjournal = {Phys. Status Solidi RRL}, - title = {Structure and {{Dynamics}} of {{Supercooled Liquid Ge}} {\textsubscript{2}} {{Sb}} {\textsubscript{2}} {{Te}} {\textsubscript{5}} from {{Machine}}‐{{Learning}}‐{{Driven Simulations}}}, - url = {https://onlinelibrary.wiley.com/doi/10.1002/pssr.202000403}, - urldate = {2021-08-10}, - volume = {15} -} - -@article{zhuanDiscriminatingHighPressureWater2020, - abstract = {Recent discoveries of dynamic ice VII and superionic ice highlight the importance of ionic diffusions in discriminating high-pressure (P) water phases. The rare event nature and the chemical bond breaking associated with these diffusions, however, make extensive simulations of these processes unpractical to ab initio and inappropriate for force field based methods. Using a first-principles neural network potential, we performed a theoretical study of water at 5-70 GPa and 300-3000 K. Long-time dynamics of protons and oxygens were found indispensable in discriminating several subtle states of water, characterized by proton's and oxygen ion's diffusion coefficients and the distribution of proton's displacements. Within dynamic ice VII, two types of proton transfer mechanisms, i.e., translational and rotational transfers, were identified to discriminate this region further into dynamic ice VII T and dynamic ice VII R. The triple point between ice VII, superionic ice (SI), and liquid exists because the loosening of the bcc oxygen skeleton is prevented by the decrease of interatomic distances at high P's. The melting of ice VII above similar to 40 GPa can be understood as a process of two individual steps: the melting of protons and the retarded melting of oxygens, responsible for the forming of SI. The boundary of the dynamic ice VII and SI lies on the continuation line ice VII's melting curve at low P's. Based on these, a detailed phase diagram is given, which may shed light on studies of water under P's in a wide range of interdisciplinary sciences.}, - annotation = {WOS:000532319600001}, - author = {Zhuan, Lin and Ye, Qijun and Pan, Ding and Li, Xin-Zheng}, - date = {2020-04-13}, - doi = {10.1088/0256-307X/37/4/043101}, - issn = {0256-307X}, - journaltitle = {Chinese Physics Letters}, - langid = {english}, - location = {{Bristol}}, - number = {4}, - pages = {043101}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Chin. Phys. Lett.}, - title = {Discriminating {{High}}-{{Pressure Water Phases Using Rare}}-{{Event Determined Ionic Dynamical Properties}}*}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {37} -} - -@article{zubatiukDevelopmentMultimodalMachine2021, - abstract = {ConspectusMachine learning interatomic potentials (MLIPs) are widely used for describing molecular energy and continue bridging the speed and accuracy gap between quantum mechanical (QM) and classical approaches like force fields. In this Account, we focus on the out-of-the-box approaches to developing transferable MLIPs for diverse chemical tasks. First, we introduce the "Accurate Neural Network engine for Molecular Energies,"ANAKIN-ME, method (or ANI for short). The ANI model utilizes Justin Smith Symmetry Functions (JSSFs) and realizes training for vast data sets. The training data set of several orders of magnitude larger than before has become the key factor of the knowledge transferability and flexibility of MLIPs. As the quantity, quality, and types of interactions included in the training data set will dictate the accuracy of MLIPs, the task of proper data selection and model training could be assisted with advanced methods like active learning (AL), transfer learning (TL), and multitask learning (MTL).Next, we describe the AIMNet "Atoms-in-Molecules Network"that was inspired by the quantum theory of atoms in molecules. The AIMNet architecture lifts multiple limitations in MLIPs. It encodes long-range interactions and learnable representations of chemical elements. We also discuss the AIMNet-ME model that expands the applicability domain of AIMNet from neutral molecules toward open-shell systems. The AIMNet-ME encompasses a dependence of the potential on molecular charge and spin. It brings ML and physical models one step closer, ensuring the correct molecular energy behavior over the total molecular charge.We finally describe perhaps the simplest possible physics-aware model, which combines ML and the extended Hückel method. In ML-EHM, "Hierarchically Interacting Particle Neural Network,"HIP-NN generates the set of a molecule- and environment-dependent Hamiltonian elements α𝜇𝜇 and K‡. As a test example, we show how in contrast to traditional Hückel theory, ML-EHM correctly describes orbital crossing with bond rotations. Hence it learns the underlying physics, highlighting that the inclusion of proper physical constraints and symmetries could significantly improve ML model generalization.}, - author = {Zubatiuk, Tetiana and Isayev, Olexandr}, - date = {2021-04}, - doi = {10.1021/acs.accounts.0c00868}, - journaltitle = {Accounts of Chemical Research}, - number = {7}, - pages = {1575--1585}, - publisher = {{American Chemical Society}}, - title = {Development of Multimodal Machine Learning Potentials: {{Toward}} a Physics-Aware Artificial Intelligence}, - volume = {54} -} - -@article{zubatiukMachineLearnedHuckel2021, - abstract = {The Hückel Hamiltonian is an incredibly simple tight-binding model famed for its ability to capture qualitative physics phenomena arising from electron interactions in molecules and materials. Part of its simplicity arises from using only two types of empirically fit physics-motivated parameters: the first describes the orbital energies on each atom and the second describes electronic interactions and bonding between atoms. By replacing these traditionally static parameters with dynamically predicted values, we vastly increase the accuracy of the extended Hückel model. The dynamic values are generated with a deep neural network, which is trained to reproduce orbital energies and densities derived from density functional theory. The resulting model retains interpretability while the deep neural network parameterization is smooth, accurate, and reproduces insightful features of the original static parameterization. Finally, we demonstrate that the Hückel model, and not the deep neural network, is responsible for capturing intricate orbital interactions in two molecular case studies. Overall, this work shows the promise of utilizing machine learning to formulate simple, accurate, and dynamically parameterized physics models.}, - author = {Zubatiuk, Tetiana and Nebgen, Benjamin and Lubbers, Nicholas and Smith, Justin S. and Zubatyuk, Roman and Zhou, Guoqing and Koh, Christopher and Barros, Kipton and Isayev, Olexandr and Tretiak, Sergei}, - date = {2021}, - doi = {10.1063/5.0052857}, - journaltitle = {The Journal of Chemical Physics}, - number = {24}, - pages = {244108}, - publisher = {{AIP Publishing LLC}}, - shorttitle = {Machine Learned {{Hückel}} Theory}, - title = {Machine Learned {{Hückel}} Theory: {{Interfacing}} Physics and Deep Neural Networks}, - volume = {154} -} - -@article{zuoPerformanceCostAssessment2020, - abstract = {Machine learning of the quantitative relationship between local environment descriptors and the potential energy surface of a system of atoms has emerged as a new frontier in the development of interatomic potentials (IAPs). Here, we present a comprehensive evaluation of machine learning IAPs (ML-IAPs) based on four local environment descriptors-atom-centered symmetry functions (ACSF), smooth overlap of atomic positions (SOAP), the spectral neighbor analysis potential (SNAP) bispectrum components, and moment tensors-using a diverse data set generated using high-throughput density functional theory (DFT) calculations. The data set comprising bcc (Li, Mo) and fcc (Cu, Ni) metals and diamond group IV semiconductors (Si, Ge) is chosen to span a range of crystal structures and bonding. All descriptors studied show excellent performance in predicting energies and forces far surpassing that of classical IAPs, as well as predicting properties such as elastic constants and phonon dispersion curves. We observe a general trade-off between accuracy and the degrees of freedom of each model and, consequently, computational cost. We will discuss these trade-offs in the context of model selection for molecular dynamics and other applications.}, - annotation = {WOS:000510531200016}, - author = {Zuo, Yunxing and Chen, Chi and Li, Xiangguo and Deng, Zhi and Chen, Yiming and Behler, Joerg and Csanyi, Gabor and Shapeev, Alexander and Thompson, Aidan P. and Wood, Mitchell A. and Ong, Shyue Ping}, - date = {2020-01-30}, - doi = {10.1021/acs.jpca.9b08723}, - issn = {1089-5639}, - journaltitle = {Journal of Physical Chemistry A}, - langid = {english}, - location = {{Washington}}, - number = {4}, - pages = {731--745}, - publisher = {{Amer Chemical Soc}}, - shortjournal = {J. Phys. Chem. A}, - title = {Performance and {{Cost Assessment}} of {{Machine Learning Interatomic Potentials}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {124} -} - - -60 other papers without abstract. - -@article{Aitken, - author = {Aitken, ZH and Sorkin, V and Yu, ZG and Chen, S and Wu, Z and B, YW Zhang - Physical Review and 2021, undefined}, - journaltitle = {APS}, - title = {Modified Embedded-Atom Method Potentials for the Plasticity and Fracture Behaviors of Unary Fcc Metals}, - url = {https://journals.aps.org/prb/abstract/10.1103/PhysRevB.103.094116} -} - -@article{ArXiv:2009.145961920, - archiveprefix = {arXiv}, - arxivid = {2009.14596v1}, - author = {arXiv preprint ArXiv:2009.14596, E Weinan - and 2020, undefined}, - date = {1920}, - eprint = {2009.14596v1}, - eprinttype = {arxiv}, - journaltitle = {arxiv.org}, - options = {useprefix=true}, - title = {Machine Learning and Computational Mathematics}, - url = {https://arxiv.org/abs/2009.14596} -} - -@article{boResearchMicrostructurePhysical2021, - author = {Bo, YANG and Guimin, L. U.}, - date = {2021}, - journaltitle = {华东理工大学学报 (自然科学版)}, - pages = {1--11}, - publisher = {{华东理工大学学报 (自然科学版)}}, - title = {Research on {{Microstructure}} and {{Physical Properties}} of {{Molten Carbonate Salt}} Based on {{Machine Learning}}} -} - -@article{Chen2021b, - author = {Chen, Zhantao and Andrejevic, Nina and Drucker, Nathan C. and Nguyen, Thanh and Xian, R. Patrick and Smidt, Tess and Wang, Yao and Ernstorfer, Ralph and Tennant, D. Alan and Chan, Maria and Li, Mingda}, - date = {2021-09}, - doi = {10.1063/5.0049111}, - journaltitle = {Chemical Physics Reviews}, - number = {3}, - pages = {031301}, - title = {Machine Learning on Neutron and X-Ray Scattering and Spectroscopies}, - url = {https://aip.scitation.org/doi/10.1063/5.0049111}, - volume = {2} -} - -@article{chenDeepLearningNonadiabatic2018, - author = {Chen, Wen-Kai and Liu, Xiang-Yang and Fang, Wei-Hai and Dral, Pavlo O. and Cui, Ganglong}, - date = {2018}, - doi = {10.1021/acs.jpclett.8b03026}, - journaltitle = {The journal of physical chemistry letters}, - number = {23}, - pages = {6702--6708}, - publisher = {{ACS Publications}}, - title = {Deep Learning for Nonadiabatic Excited-State Dynamics}, - volume = {9} -} - -@article{chengBuildingMachineLearning2021, - author = {Cheng, Zheng and Du, Jiahui and Zhang, Lei and Ma, Jing and Li, Wei and Li, Shuhua}, - date = {2021}, - title = {Building {{Machine Learning Force Fields}} of {{Proteins}} with {{Fragment}}-{{Based Approach}} and {{Transfer Learning}}} -} - -@article{chenStudyOpticalPhonon2021, - author = {Chen, Wei and Li, Liang-Sheng}, - date = {2021}, - doi = {10.1063/5.0049464}, - journaltitle = {Journal of Applied Physics}, - number = {24}, - pages = {244104}, - publisher = {{AIP Publishing LLC}}, - title = {The Study of the Optical Phonon Frequency of {{3C}}-{{SiC}} by Molecular Dynamics Simulations with Deep Neural Network Potential}, - volume = {129} -} - -@article{christensenRoleGradientsMachine2020, - author = {Christensen, Anders S. and von Lilienfeld, O. Anatole}, - date = {2020}, - doi = {10.1088/2632-2153/abba6f}, - journaltitle = {Machine Learning: Science and Technology}, - number = {4}, - options = {useprefix=true}, - pages = {045018}, - publisher = {{IOP Publishing}}, - title = {On the Role of Gradients for Machine Learning of Molecular Energies and Forces}, - volume = {1} -} - -@article{chuLonglivedHotElectron2020, - author = {Chu, Weibin and Saidi, Wissam A. and Prezhdo, Oleg V.}, - date = {2020}, - doi = {10.1021/acsnano.0c04736}, - journaltitle = {ACS nano}, - number = {8}, - pages = {10608--10615}, - publisher = {{ACS Publications}}, - shorttitle = {Long-Lived Hot Electron in a Metallic Particle for Plasmonics and Catalysis}, - title = {Long-Lived Hot Electron in a Metallic Particle for Plasmonics and Catalysis: {{Ab}} Initio Nonadiabatic Molecular Dynamics with Machine Learning}, - volume = {14} -} - -@article{desaiImplementingNeuralNetwork2020, - archiveprefix = {arXiv}, - author = {Desai, Saaketh and Reeve, Samuel Temple and Belak, James F.}, - date = {2020}, - eprint = {2002.00054}, - eprinttype = {arxiv}, - title = {Implementing a Neural Network Interatomic Model with Performance Portability for Emerging Exascale Architectures} -} - -@article{dralNonadiabaticExcitedstateDynamics2018, - author = {Dral, Pavlo O. and Barbatti, Mario and Thiel, Walter}, - date = {2018}, - doi = {10.1021/acs.jpclett.8b02469}, - journaltitle = {The journal of physical chemistry letters}, - number = {19}, - pages = {5660--5663}, - publisher = {{ACS Publications}}, - title = {Nonadiabatic Excited-State Dynamics with Machine Learning}, - volume = {9} -} - -@article{eMachineLearningComputational2020, - archiveprefix = {arXiv}, - author = {E, Weinan}, - date = {2020}, - eprint = {2009.14596}, - eprinttype = {arxiv}, - title = {Machine {{Learning}} and {{Computational Mathematics}}} -} - -@report{fabrizioDeterministicStatisticalApproaches2020, - author = {Fabrizio, Alberto}, - date = {2020}, - institution = {{EPFL}}, - title = {Deterministic and {{Statistical Approaches}} to {{Quantum Chemistry}}} -} - -@article{fineganApplicationDatadrivenMethods2020, - author = {Finegan, Donal P. and Zhu, Juner and Feng, Xuning and Keyser, Matt and Ulmefors, Marcus and Li, Wei and Bazant, Martin Z. and Cooper, Samuel J.}, - date = {2020}, - journaltitle = {Joule}, - publisher = {{Elsevier}}, - title = {The Application of Data-Driven Methods and Physics-Based Learning for Improving Battery Safety} -} - -@article{grasselliHeatChargeTransport2020, - author = {Grasselli, Federico and Stixrude, Lars and Baroni, Stefano}, - date = {2020}, - doi = {10.1038/s41467-020-17275-5}, - journaltitle = {Nature communications}, - number = {1}, - pages = {1--7}, - publisher = {{Nature Publishing Group}}, - title = {Heat and Charge Transport in {{H}} 2 {{O}} at Ice-Giant Conditions from Ab Initio Molecular Dynamics Simulations}, - volume = {11} -} - -@article{grisafiTransferableMachinelearningModel2018, - author = {Grisafi, Andrea and Fabrizio, Alberto and Meyer, Benjamin and Wilkins, David M. and Corminboeuf, Clemence and Ceriotti, Michele}, - date = {2018}, - doi = {10.1021/acscentsci.8b00551}, - journaltitle = {ACS central science}, - number = {1}, - pages = {57--64}, - publisher = {{ACS Publications}}, - title = {Transferable Machine-Learning Model of the Electron Density}, - volume = {5} -} - -@article{guenzaAccuracyTransferabilityEfficiency2018, - author = {Guenza, M. G. and Dinpajooh, M. and McCarty, J. and Lyubimov, I. Y.}, - date = {2018}, - doi = {10.1021/acs.jpcb.8b06687}, - journaltitle = {The Journal of Physical Chemistry B}, - number = {45}, - pages = {10257--10278}, - publisher = {{ACS Publications}}, - title = {Accuracy, Transferability, and Efficiency of Coarse-Grained Models of Molecular Liquids}, - volume = {122} -} - -@article{kobayashiHighthroughputProductionForcefields2020, - author = {Kobayashi, Ryo and Miyaji, Yasuhiro and Nakano, Koki and Nakayama, Masanobu}, - date = {2020}, - doi = {10.1063/5.0015373}, - journaltitle = {APL Materials}, - number = {8}, - pages = {081111}, - publisher = {{AIP Publishing LLC}}, - title = {High-Throughput Production of Force-Fields for Solid-State Electrolyte Materials}, - volume = {8} -} - -@article{koEnablingLargeScaleCondensedPhase2020, - archiveprefix = {arXiv}, - author = {Ko, Hsin-Yu and Santra, Biswajit and DiStasio Jr, Robert A.}, - date = {2020}, - eprint = {2011.07209}, - eprinttype = {arxiv}, - shorttitle = {Enabling {{Large}}-{{Scale Condensed}}-{{Phase Hybrid Density Functional Theory Based}} \$ {{Ab}} \$\$ {{Initio}} \$ {{Molecular Dynamics II}}}, - title = {Enabling {{Large}}-{{Scale Condensed}}-{{Phase Hybrid Density Functional Theory Based}} \$ {{Ab}} \$\$ {{Initio}} \$ {{Molecular Dynamics II}}: {{Extensions}} to the {{Isobaric}}-{{Isoenthalpic}} and {{Isobaric}}-{{Isothermal Ensembles}}} -} - -@article{Liang, - author = {Liang, W and Lu, G and Yu, J}, - date = {2020-11-02}, - journaltitle = {Wiley Online Library}, - title = {Molecular Dynamics Simulations of Molten Magnesium Chloride Using {{Machine}}‐{{Learning}}‐{{Based}} Deep Potential}, - url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/adts.202000180} -} - -@article{liDeepNeuralNetwork2020, - author = {Li, Ruiyang and Liu, Zeyu and Rohskopf, Andrew and Gordiz, Kiarash and Henry, Asegun and Lee, Eungkyu and Luo, Tengfei}, - date = {2020}, - doi = {10.1063/5.0025051}, - journaltitle = {Applied Physics Letters}, - number = {15}, - pages = {152102}, - publisher = {{AIP Publishing LLC}}, - title = {A Deep Neural Network Interatomic Potential for Studying Thermal Conductivity of β-{{Ga2O3}}}, - volume = {117} -} - -@article{liEffectsDensityComposition2020, - author = {Li, Wenwen and Ando, Yasunobu and Watanabe, Satoshi}, - date = {2020}, - doi = {10.1063/5.0026289}, - journaltitle = {The Journal of Chemical Physics}, - number = {16}, - pages = {164119}, - publisher = {{AIP Publishing LLC}}, - shorttitle = {Effects of Density and Composition on the Properties of Amorphous Alumina}, - title = {Effects of Density and Composition on the Properties of Amorphous Alumina: {{A}} High-Dimensional Neural Network Potential Study}, - volume = {153} -} - -@article{Lin2020, - author = {Lin, Qidong and Zhang, Yaolong and Zhao, Bin and Jiang, Bin}, - date = {2020-04}, - doi = {10.1063/5.0004944}, - number = {15}, - publisher = {{American Institute of Physics Inc.}}, - title = {Automatically Growing Global Reactive Neural Network Potential Energy Surfaces: {{A}} Trajectory-Free Active Learning Strategy}, - url = {https://aip.scitation.org/doi/abs/10.1063/5.0004944}, - volume = {152} -} - -@article{Lindsey2020, - author = {Lindsey, Rebecca K. RK and Fried, LE Laurence E. and …, N Goldman - The Journal of Chemical and 2020, undefined and Goldman, Nir and Bastea, Sorin}, - date = {2020-10}, - doi = {10.1063/5.0021965}, - number = {13}, - publisher = {{American Institute of Physics Inc.}}, - title = {Active Learning for Robust, High-Complexity Reactive Atomistic Simulations}, - url = {https://aip.scitation.org/doi/abs/10.1063/5.0021965}, - volume = {153} -} - -@article{luFutureDirectionsChemical2021, - author = {Lu, Yuyuan and Deng, Geng and Shuai, Zhigang}, - date = {2021}, - doi = {10.1515/pac-2020-1006}, - journaltitle = {Pure and Applied Chemistry}, - publisher = {{De Gruyter}}, - title = {Future Directions of Chemical Theory and Computation} -} - -@article{luUniversalApproximationTheorem2020, - archiveprefix = {arXiv}, - author = {Lu, Yulong and Lu, Jianfeng}, - date = {2020}, - eprint = {2004.08867}, - eprinttype = {arxiv}, - title = {A Universal Approximation Theorem of Deep Neural Networks for Expressing Probability Distributions} -} - -@article{moradzadehUnderstandingSimpleLiquids2021, - author = {Moradzadeh, A. and Aluru, N. R.}, - date = {2021}, - doi = {10.1063/5.0046226}, - journaltitle = {The Journal of Chemical Physics}, - number = {20}, - pages = {204503}, - publisher = {{AIP Publishing LLC}}, - title = {Understanding Simple Liquids through Statistical and Deep Learning Approaches}, - volume = {154} -} - -@article{Mortensen2020, - archiveprefix = {arXiv}, - arxivid = {2007.07523v1}, - author = {Mortensen, HL Henrik Lund and Meldgaard, Søren Ager SA and Bisbo, Malthe Kjær and Christiansen, Mads Peter V. and Hammer, Bjørk and B, MK Bisbo - Physical Review and 2020, undefined}, - date = {2020-08}, - doi = {10.1103/physrevb.102.075427}, - eprint = {2007.07523v1}, - eprinttype = {arxiv}, - number = {7}, - publisher = {{American Physical Society}}, - title = {Atomistic Structure Learning Algorithm with Surrogate Energy Model Relaxation}, - url = {https://journals.aps.org/prb/abstract/10.1103/PhysRevB.102.075427}, - volume = {102} -} - -@article{networkMachineLearningNanoScale, - author = {Network, BPN Behler-Parrinello}, - title = {Machine {{Learning}} in {{Nano}}-{{Scale Biomedical Engineering}}} -} - -@article{Novikov2019, - archiveprefix = {arXiv}, - arxivid = {1909.06244v3}, - author = {Novikov, IS Ivan S. and Shapeev, Alexander V. and Suleimanov, Yury V. and of chemical …, AV Shapeev - The Journal and 2019, undefined}, - date = {2019-12}, - doi = {10.1063/1.5127561}, - eprint = {1909.06244v3}, - eprinttype = {arxiv}, - number = {22}, - options = {useprefix=true}, - publisher = {{American Institute of Physics Inc.}}, - title = {Ring Polymer Molecular Dynamics and Active Learning of Moment Tensor Potential for Gas-Phase Barrierless Reactions: {{Application}} to {{S}} + {{H2}}}, - url = {https://aip.scitation.org/doi/abs/10.1063/1.5127561}, - volume = {151} -} - -@article{novikovAutomatedCalculationThermal2018, - author = {Novikov, Ivan S. and Suleimanov, Yury V. and Shapeev, Alexander V.}, - date = {2018}, - doi = {10.1039/C8CP06037A}, - journaltitle = {Physical Chemistry Chemical Physics}, - number = {46}, - pages = {29503--29512}, - publisher = {{Royal Society of Chemistry}}, - title = {Automated Calculation of Thermal Rate Coefficients Using Ring Polymer Molecular Dynamics and Machine-Learning Interatomic Potentials with Active Learning}, - volume = {20} -} - -@inproceedings{oehmckeModelingRutileTiO1102020, - author = {Oehmcke, Stefan and Teusch, Thomas and Petersen, Thorben and Klüner, Thorsten and Kramer, Oliver}, - booktitle = {2020 {{International Joint Conference}} on {{Neural Networks}} ({{IJCNN}})}, - date = {2020}, - doi = {10.1109/IJCNN48605.2020.9207275}, - pages = {1--7}, - publisher = {{IEEE}}, - title = {Modeling {{H}} 2 {{O}}/{{Rutile}}-{{TiO}} 2 (110) {{Potential Energy Surfaces}} with {{Deep Networks}}} -} - -@inproceedings{phuongDeepLearningInteratomic2020, - author = {Phuong, H. S. M. and Starostenkov, M. D. and Trung, N. T. H.}, - booktitle = {Эволюция Дефектных Структур в Конденсированных Средах}, - date = {2020}, - pages = {141--142}, - title = {Deep Learning Interatomic Potential for Simulation of Radiation Damage in Vanadium-Rich {{V}}-{{Cr}}-{{Ti}} Ternary Alloys} -} - -@article{Pun2020, - archiveprefix = {arXiv}, - arxivid = {2009.06533v3}, - author = {Pun, GPP P.Purja and Yamakov, V. and Hickman, J. and Glaessgen, E. H. and Mishin, Y. and …, EH Glaessgen - Physical Review and 2020, undefined}, - date = {2020-11}, - doi = {10.1103/physrevmaterials.4.113807}, - eprint = {2009.06533v3}, - eprinttype = {arxiv}, - number = {11}, - publisher = {{American Physical Society}}, - title = {Development of a General-Purpose Machine-Learning Interatomic Potential for Aluminum by the Physically Informed Neural Network Method}, - url = {https://journals.aps.org/prmaterials/abstract/10.1103/PhysRevMaterials.4.113807}, - volume = {4} -} - -@article{Reviews, - author = {Reviews, J Behler - Chemical and 2021, undefined}, - journaltitle = {ACS Publications}, - title = {Four Generations of High-Dimensional Neural Network Potentials}, - url = {https://pubs.acs.org/doi/abs/10.1021/acs.chemrev.0c00868} -} - -@article{samantaRepresentingLocalAtomic2018, - author = {Samanta, Amit}, - date = {2018}, - doi = {10.1063/1.5055772}, - journaltitle = {The Journal of chemical physics}, - number = {24}, - pages = {244102}, - publisher = {{AIP Publishing LLC}}, - title = {Representing Local Atomic Environment Using Descriptors Based on Local Correlations}, - volume = {149} -} - -@article{Science, - author = {Science, WF Reinhart - Computational Materials and 2021, undefined}, - journaltitle = {Elsevier}, - title = {Unsupervised Learning of Atomic Environments from Simple Features}, - url = {https://www.sciencedirect.com/science/article/pii/S0927025621002366} -} - -@article{Shaidu, - author = {Shaidu, Y and Küçükbenli, E and Lot, R and Pellegrini, F}, - journaltitle = {nature.com}, - title = {A Systematic Approach to Generating Accurate Neural Network Potentials: The Case of Carbon}, - url = {https://www.nature.com/articles/s41524-021-00508-6} -} - -@article{Shapeev, - author = {Shapeev, AV and Podryabinkin, EV and Gubaev, K}, - journaltitle = {iopscience.iop.org}, - title = {Elinvar Effect in {{β}}-{{Ti}} Simulated by on-the-Fly Trained Moment Tensor Potential}, - url = {https://iopscience.iop.org/article/10.1088/1367-2630/abc392/meta} -} - -@article{singraberLibrarybasedLAMMPSImplementation2019, - author = {Singraber, Andreas and Behler, Jörg and Dellago, Christoph}, - date = {2019}, - doi = {10.1021/acs.jctc.8b00770}, - journaltitle = {Journal of chemical theory and computation}, - number = {3}, - pages = {1827--1840}, - publisher = {{ACS Publications}}, - title = {Library-Based {{LAMMPS}} Implementation of High-Dimensional Neural Network Potentials}, - volume = {15} -} - -@article{Sivaramana, - author = {Sivaraman, G and Krishnamoorthy, AN and …, M Baur - npj Computational and 2020, undefined}, - journaltitle = {nature.com}, - title = {Machine-Learned Interatomic Potentials by Active Learning: Amorphous and Liquid Hafnium Dioxide}, - url = {https://www.nature.com/articles/s41524-020-00367-7} -} - -@article{Smith, - author = {Smith, JS and Nebgen, B and Mathew, N and Chen, J}, - journaltitle = {nature.com}, - title = {Automated Discovery of a Robust Interatomic Potential for Aluminum}, - url = {https://www.nature.com/articles/s41467-021-21376-0} -} - -@article{Tian2021, - archiveprefix = {arXiv}, - arxivid = {2010.06896v1}, - author = {Tian, Yuan and Xue, Dezhen and Yuan, Ruihao and Zhou, Yumei and Ding, Xiangdong and Sun, Jun and Lookman, Turab and …, J Sun - Physical Review and 2021, undefined}, - date = {2021-01}, - doi = {10.1103/physrevmaterials.5.013802}, - eprint = {2010.06896v1}, - eprinttype = {arxiv}, - number = {1}, - publisher = {{American Physical Society}}, - title = {Efficient Estimation of Material Property Curves and Surfaces via Active Learning}, - url = {https://journals.aps.org/prmaterials/abstract/10.1103/PhysRevMaterials.5.013802}, - volume = {5} -} - -@article{townshendGeneralizableProteinInterface2018, - archiveprefix = {arXiv}, - author = {Townshend, R. J. and Bedi, Rishi and Dror, Ron O.}, - date = {2018}, - eprint = {1807.01297}, - eprinttype = {arxiv}, - title = {Generalizable Protein Interface Prediction with End-to-End Learning} -} - -@article{tsubakiFastAccurateMolecular2018, - author = {Tsubaki, Masashi and Mizoguchi, Teruyasu}, - date = {2018}, - doi = {10.1021/acs.jpclett.8b01837}, - journaltitle = {The journal of physical chemistry letters}, - number = {19}, - pages = {5733--5741}, - publisher = {{ACS Publications}}, - shorttitle = {Fast and Accurate Molecular Property Prediction}, - title = {Fast and Accurate Molecular Property Prediction: Learning Atomic Interactions and Potentials with Neural Networks}, - volume = {9} -} - -@article{vanspeybroeckModelingSpatiotemporalProcesses2021, - author = {Van Speybroeck, Veronique and Vandenhaute, Sander and Hoffman, Alexander EJ and Rogge, Sven MJ}, - date = {2021}, - doi = {10.1016/j.trechm.2021.04.003}, - journaltitle = {Trends in Chemistry}, - publisher = {{Elsevier}}, - title = {Towards Modeling Spatiotemporal Processes in Metal–Organic Frameworks} -} - -@article{Wen, - author = {Wen, M and Tadmor, EB}, - date = {2020-08-14}, - journaltitle = {nature.com}, - title = {Uncertainty Quantification in Molecular Simulations with Dropout Neural Network Potentials}, - url = {https://www.nature.com/articles/s41524-020-00390-8} -} - -@article{westermayrDeepLearningUV2020, - author = {Westermayr, Julia and Marquetand, Philipp}, - date = {2020}, - doi = {10.1063/5.0021915}, - journaltitle = {The Journal of Chemical Physics}, - number = {15}, - pages = {154112}, - publisher = {{AIP Publishing LLC}}, - shorttitle = {Deep Learning for {{UV}} Absorption Spectra with {{SchNarc}}}, - title = {Deep Learning for {{UV}} Absorption Spectra with {{SchNarc}}: {{First}} Steps toward Transferability in Chemical Compound Space}, - volume = {153} -} - -@article{westermayrMachineLearningNonadiabatic2020, - author = {Westermayr, Julia and Marquetand, Philipp}, - date = {2020}, - doi = {10.1039/9781839160233-00076}, - journaltitle = {Machine Learning in Chemistry}, - pages = {76}, - publisher = {{Royal Society of Chemistry}}, - title = {Machine Learning for Nonadiabatic Molecular Dynamics}, - volume = {17} -} - -@article{willattDatadrivenConstructionPeriodic2018, - archiveprefix = {arXiv}, - author = {Willatt, Michael J. and Musil, Félix and Ceriotti, Michele}, - date = {2018}, - eprint = {1807.00236}, - eprinttype = {arxiv}, - title = {A Data-Driven Construction of the Periodic Table of the Elements} -} - -@article{willattTheoryPracticeAtomdensity2018, - author = {Willatt, Michael J. and Musil, Félix and Ceriotti, Michele}, - date = {2018}, - journaltitle = {arXiv preprint}, - title = {Theory and Practice of Atom-Density Representations for Machine Learning} -} - -@article{xuModelingPredictingResponses2018, - author = {Xu, Ben and Nan, Ce-Wen}, - date = {2018}, - doi = {10.1557/mrs.2018.259}, - journaltitle = {MRS Bulletin}, - number = {11}, - pages = {829--833}, - publisher = {{Cambridge University Press}}, - title = {Modeling and Predicting Responses of Magnetoelectric Materials}, - volume = {43} -} - -@article{yangTheoreticalInvestigationHalide2020, - abstract = {The solar cell based on organic-inorganic hybrid halide perovskite is progressing amazingly fast in last decade owing to the robust experimental and theoretical investigations. First-principles calculation is one of the crucial ways to understand the nature of the materials and is practically helpful to the development and application of perovskite solar cells. Here, we briefly review the progress of theoretical studies we made in the last few years on the modification of electronic structures of perovskites by varying the composition, configuration, and structure, and the new understandings into the defect properties of halide perovskites for solar cell and optoelectronic applications. These understandings are foundations and new starting points for future investigations. We hope the experience and inspiration gained from these studies encourage more theoretical explorations for new functional perovskite-based materials.}, - author = {Yang, Jingxiu and Zhang, Peng and Wang, Jianping and Wei, Su Huai}, - date = {2020-10}, - doi = {10.1088/1674-1056/abb3f6}, - journaltitle = {Chinese Physics B}, - number = {10}, - publisher = {{IOP Publishing Ltd}}, - title = {Theoretical Investigation of Halide Perovskites for Solar Cell and Optoelectronic Applications}, - volume = {29} -} - -@article{Yu2020, - author = {Yu, Haijun and Tian, Xinyuan and arXiv preprint ArXiv:2009.02327, Q Li - and 2020, undefined and E, Weinan and Li, Qianxiao}, - date = {2020-09}, - journaltitle = {arxiv.org}, - options = {useprefix=true}, - title = {{{OnsagerNet}}: {{Learning}} Stable and Interpretable Dynamics Using a Generalized Onsager Principle}, - url = {http://arxiv.org/abs/2009.02327 http://dx.doi.org/10.1017/S1743921320000629 https://arxiv.org/abs/2009.02327} -} - -@article{Zaverkin, - author = {Zaverkin, V and Kästner, J}, - date = {2021-05-12}, - journaltitle = {iopscience.iop.org}, - title = {Exploration of Transferable and Uniformly Accurate Neural Network Interatomic Potentials Using Optimal Experimental Design}, - url = {https://iopscience.iop.org/article/10.1088/2632-2153/abe294/meta} -} - -@article{zhaiDiscoveryDesignSoft2020, - author = {Zhai, Chenxi and Li, Tianjiao and Shi, Haoyuan and Yeo, Jingjie}, - date = {2020}, - doi = {10.1039/D0TB00896F}, - journaltitle = {Journal of Materials Chemistry B}, - number = {31}, - pages = {6562--6587}, - publisher = {{Royal Society of Chemistry}}, - title = {Discovery and Design of Soft Polymeric Bio-Inspired Materials with Multiscale Simulations and Artificial Intelligence}, - volume = {8} -} - -@article{zhaiInferringMicrobubbleDynamics2021, - archiveprefix = {arXiv}, - author = {Zhai, Hanfeng and Hu, Guohui}, - date = {2021}, - eprint = {2105.07179}, - eprinttype = {arxiv}, - title = {Inferring Micro-Bubble Dynamics with Physics-Informed Deep Learning} -} - -@article{zhangArrestedPhaseSeparation2021a, - abstract = {We present large-scale dynamical simulations of electronic phase separation in the single-band double-exchange model based on deep-learning neural-network potentials trained from small-size exact diagonalization solutions. We uncover an intriguing correlation-induced freezing behavior as doped holes are segregated from half-filled insulating background during equilibration. While the aggregation of holes is stabilized by the formation of ferromagnetic clusters through Hund's coupling between charge carriers and local magnetic moments, this stabilization also creates confining potentials for holes when antiferromagnetic spin-spin correlation is well developed in the background. The dramatically reduced mobility of the self-trapped holes prematurely disrupts further growth of the ferromagnetic clusters, leading to an arrested phase separation. Implications of our findings for phase separation dynamics in materials that exhibit colossal magnetoresistance effect are discussed.}, - archiveprefix = {arXiv}, - author = {Zhang, Puhan and Chern, Gia-Wei}, - date = {2021-05-17}, - eprint = {2105.08221}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {cond-mat}, - shorttitle = {Arrested Phase Separation in Double-Exchange Models}, - title = {Arrested Phase Separation in Double-Exchange Models: Machine-Learning Enabled Large-Scale Simulation}, - url = {http://arxiv.org/abs/2105.08221}, - urldate = {2021-08-10} -} - -@article{zhangPhysicallyInspiredAtomcentered2021, - author = {Zhang, Kangyu and Yin, Lichang and Liu, Gang}, - date = {2021}, - doi = {10.1016/j.commatsci.2020.110071}, - journaltitle = {Computational Materials Science}, - pages = {110071}, - publisher = {{Elsevier}}, - title = {Physically Inspired Atom-Centered Symmetry Functions for the Construction of High Dimensional Neural Network Potential Energy Surfaces}, - volume = {186} -} - -@article{zhaoAdaptiveGeneticAlgorithm2020, - author = {Zhao, Xin and Wu, Shunqing and Nguyen, Manh Cuong and Ho, Kai-Ming and Wang, Cai-Zhuang}, - date = {2020}, - doi = {10.1007/978-3-319-44680-6_73}, - journaltitle = {Handbook of Materials Modeling: Applications: Current and Emerging Materials}, - pages = {2757--2776}, - publisher = {{Springer}}, - title = {Adaptive {{Genetic Algorithm}} for {{Structure Prediction}} and {{Application}} to {{Magnetic Materials}}} -} - diff --git a/source/_data/reviews.bib b/source/_data/reviews.bib deleted file mode 100644 index 1ffb1cf1..00000000 --- a/source/_data/reviews.bib +++ /dev/null @@ -1,715 +0,0 @@ -@thesis{balasubramanianDiscoveryImplementationFast2019, - author = {Balasubramanian, Adarsh}, - date = {2019}, - institution = {{Johns Hopkins University}}, - title = {Discovery and {{Implementation}} of Fast, Accurate and Transferable {{Many}}-Body {{Interatomic Potentials}}}, - type = {PhD Thesis} -} - -@article{behlerFourGenerationsHighdimensional2021, - abstract = {Since their introduction about 25 years ago, machine learning (ML) potentials have become an important tool in the field of atomistic simulations. After the initial decade, in which neural networks were successfully used to construct potentials for rather small molecular systems, the development of high-dimensional neural network potentials (HDNNPs) in 2007 opened the way for the application of ML potentials in simulations of large systems containing thousands of atoms. To date, many other types of ML potentials have been proposed continuously increasing the range of problems that can be studied. In this review, the methodology of the family of HDNNPs including new recent developments will be discussed using a classification scheme into four generations of potentials, which is also applicable to many other types of ML potentials. The first generation is formed by early neural network potentials designed for low-dimensional systems. High-dimensional neural network potentials established the second generation and are based on three key steps: first, the expression of the total energy as a sum of environment-dependent atomic energy contributions; second, the description of the atomic environments by atom-centered symmetry functions as descriptors fulfilling the requirements of rotational, translational, and permutation invariance; and third, the iterative construction of the reference electronic structure data sets by active learning. In third-generation HDNNPs, in addition, long-range interactions are included employing environment-dependent partial charges expressed by atomic neural networks. In fourth-generation HDNNPs, which are just emerging, in addition, nonlocal phenomena such as long-range charge transfer can be included. The applicability and remaining limitations of HDNNPs are discussed along with an outlook at possible future developments.}, - author = {Behler, Jörg}, - date = {2021}, - doi = {10.1021/acs.chemrev.0c00868}, - journaltitle = {Chemical Reviews}, - publisher = {{American Chemical Society}}, - title = {Four Generations of High-Dimensional Neural Network Potentials}, - url = {https://pubs.acs.org/doi/abs/10.1021/acs.chemrev.0c00868} -} - -@thesis{carboneDynamicalProcessesCondensed2021, - abstract = {In this thesis, we study a broad range of physical phenomena from the perspectives of theorydriven, and machine learning models. We begin by introducing a generalization of the Momentum Average method for finding numerically exact Green’s functions of arbitrary polaron systems at zero and finite temperature. This method utilizes the physical ansatz that phonons are produced largely in clouds, and systematically constructs a closure of auxiliary Green’s functions to ultimately solve for the spectrum. We seamlessly apply this method to a variety of problems, including the Holstein, Peierls, and mixed-boson mode models. Next, we leverage fundamental quantum mechanics to develop a microscopic model of exciton and trion scattering in monolayer transition metal dichalcogenides. We conclude that elastic scattering mechanisms are largely the dominant contributor, and confirm that our calculated doping-dependent linewidths qualitatively agree with experiment. In addition, we use Monte Carlo dynamics to examine entropically activated dynamics in continuous phase space models, and show that global and local dynamics both exhibit entropy-driven activation. The second type of work discussed in this thesis pertains to data-driven machine learning models. These approaches offer the utility of instantaneous inference, which has tremendous potential application in applied science in areas such as surrogate modeling and creating digital twins of expensive experiments. First, we demonstrate that x-ray absorption spectra can be used to classify absorbing sites’ local atomic information, specifically its coordination number. Next, we show that graph-based neural networks can to quantitative accuracy, predict the x-ray absorption spectrum of small molecules in the QM9 database. We highlight the various ways in which these types of methodologies can be applied to e.g. closing the design loop and surrogate modeling in general.}, - author = {Carbone, Matthew Ralph}, - date = {2021}, - institution = {{Columbia University}}, - shorttitle = {Dynamical Processes in the Condensed Phase}, - title = {Dynamical Processes in the Condensed Phase: Methods and Models}, - type = {PhD Thesis} -} - -@article{carleoMachineLearningPhysical2019, - abstract = {Machine learning (ML) encompasses a broad range of algorithms and modeling tools used for a vast array of data processing tasks, which has entered most scientific disciplines in recent years. This article reviews in a selective way the recent research on the interface between machine learning and the physical sciences. This includes conceptual developments in ML motivated by physical insights, applications of machine learning techniques to several domains in physics, and cross fertilization between the two fields. After giving a basic notion of machine learning methods and principles, examples are described of how statistical physics is used to understand methods in ML. This review then describes applications of ML methods in particle physics and cosmology, quantum many-body physics, quantum computing, and chemical and material physics. Research and development into novel computing architectures aimed at accelerating ML are also highlighted. Each of the sections describe recent successes as well as domain-specific methodology and challenges.}, - annotation = {WOS:000505697300001}, - author = {Carleo, Giuseppe and Cirac, Ignacio and Cranmer, Kyle and Daudet, Laurent and Schuld, Maria and Tishby, Naftali and Vogt-Maranto, Leslie and Zdeborova, Lenka}, - date = {2019-12-06}, - doi = {10.1103/RevModPhys.91.045002}, - issn = {0034-6861}, - journaltitle = {Reviews of Modern Physics}, - langid = {english}, - location = {{College Pk}}, - number = {4}, - pages = {045002}, - publisher = {{Amer Physical Soc}}, - shortjournal = {Rev. Mod. Phys.}, - title = {Machine Learning and the Physical Sciences}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/5}, - urldate = {2021-08-06}, - volume = {91} -} - -@article{coleyAutonomousDiscoveryChemical2020a, - abstract = {This two-part Review examines how automation has contributed to different aspects of discovery in the chemical sciences. In this first part, we describe a classification for discoveries of physical matter (molecules, materials, devices), processes, and models and how they are unified as search problems. We then introduce a set of questions and considerations relevant to assessing the extent of autonomy. Finally, we describe many case studies of discoveries accelerated by or resulting from computer assistance and automation from the domains of synthetic chemistry, drug discovery, inorganic chemistry, and materials science. These illustrate how rapid advancements in hardware automation and machine learning continue to transform the nature of experimentation and modeling. Part two reflects on these case studies and identifies a set of open challenges for the field.}, - annotation = {WOS:000538529000001}, - author = {Coley, Connor W. and Eyke, Natalie S. and Jensen, Klavs F.}, - date = {2020-12-14}, - doi = {10.1002/anie.201909987}, - issn = {1433-7851}, - journaltitle = {Angewandte Chemie-International Edition}, - langid = {english}, - location = {{Weinheim}}, - number = {51}, - pages = {22858--22893}, - publisher = {{Wiley-V C H Verlag Gmbh}}, - shortjournal = {Angew. Chem.-Int. Edit.}, - shorttitle = {Autonomous {{Discovery}} in the {{Chemical Sciences Part I}}}, - title = {Autonomous {{Discovery}} in the {{Chemical Sciences Part I}}: {{Progress}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {59} -} - -@thesis{costamagnaDesigningModelsUsing2020, - author = {Costamagna, Andrea}, - date = {2020}, - institution = {{Politecnico di Torino}}, - shorttitle = {Designing Models Using Machine Learning}, - title = {Designing Models Using Machine Learning: One-Body Reduced Density Matrices and Spectra}, - type = {PhD Thesis} -} - -@thesis{doyleInterfacialPotentialsIon2020, - author = {Doyle, Carrie Conor}, - date = {2020}, - institution = {{University of Cincinnati}}, - title = {Interfacial {{Potentials}} in {{Ion Solvation}}}, - type = {PhD Thesis} -} - -@article{dralMolecularExcitedStates2021, - abstract = {Theoretical simulations of electronic excitations and associated processes in molecules are indispensable for fundamental research and technological innovations. However, such simulations are notoriously challenging to perform with quantum mechanical methods. Advances in machine learning open many new avenues for assisting molecular excited-state simulations. In this Review, we track such progress, assess the current state of the art and highlight the critical issues to solve in the future. We overview a broad range of machine learning applications in excited-state research, which include the prediction of molecular properties, improvements of quantum mechanical methods for the calculations of excited-state properties and the search for new materials. Machine learning approaches can help us understand hidden factors that influence photo-processes, leading to a better control of such processes and new rules for the design of materials for optoelectronic applications.}, - annotation = {WOS:000652426900001}, - author = {Dral, Pavlo O. and Barbatti, Mario}, - date = {2021-06}, - doi = {10.1038/s41570-021-00278-1}, - journaltitle = {Nature Reviews Chemistry}, - langid = {english}, - location = {{Berlin}}, - number = {6}, - pages = {388--405}, - publisher = {{Nature Research}}, - shortjournal = {Nat. Rev. Chem.}, - title = {Molecular Excited States through a Machine Learning Lens}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {5} -} - -@incollection{fischerCharacterizingMagneticSkyrmions2021, - author = {Fischer, Peter and Roy, Sujoy}, - booktitle = {Magnetic {{Skyrmions}} and {{Their Applications}}}, - date = {2021}, - pages = {55--97}, - publisher = {{Elsevier}}, - title = {Characterizing Magnetic Skyrmions at Their Fundamental Length and Time Scales} -} - -@article{glielmoUnsupervisedLearningMethods2021, - abstract = {Unsupervised learning is becoming an essential tool to analyze the increasingly large amounts of data produced by atomistic and molecular simulations, in material science, solid state physics, biophysics, and biochemistry. In this Review, we provide a comprehensive overview of the methods of unsupervised learning that have been most commonly used to investigate simulation data and indicate likely directions for further developments in the field. In particular, we discuss feature representation of molecular systems and present state-of-the-art algorithms of dimensionality reduction, density estimation, and clustering, and kinetic models. We divide our discussion into self-contained sections, each discussing a specific method. In each section, we briefly touch upon the mathematical and algorithmic foundations of the method, highlight its strengths and limitations, and describe the specific ways in which it has been used-or can be used-to analyze molecular simulation data.}, - author = {Glielmo, Aldo and Husic, Brooke E. and Rodriguez, Alex and Clementi, Cecilia and Noé, Frank and Laio, Alessandro}, - date = {2021}, - doi = {10.1021/acs.chemrev.0c01195}, - journaltitle = {Chemical Reviews}, - publisher = {{ACS Publications}}, - title = {Unsupervised {{Learning Methods}} for {{Molecular Simulation Data}}}, - url = {https://pubs.acs.org/doi/abs/10.1021/acs.chemrev.0c01195} -} - -@article{goncalvesmarquesStructureDynamicsMaterials2020, - author = {Gonçalves Marques, Mário Rui}, - date = {2020}, - title = {The Structure and Dynamics of Materials Using Machine Learning} -} - -@article{greenstreetMachinelearningassistedModeling2021, - abstract = {By integrating artificial intelligence algorithms and physics-based simulations, researchers are developing new models that are both reliable and interpretable.}, - annotation = {WOS:000668845000013}, - author = {Greenstreet, Sarah}, - date = {2021-07-01}, - doi = {10.1063/PT.3.4794}, - issn = {0031-9228}, - journaltitle = {Physics Today}, - langid = {english}, - location = {{Melville}}, - number = {7}, - pages = {42--47}, - publisher = {{Amer Inst Physics}}, - shortjournal = {Phys. Today}, - title = {Machine-Learning-Assisted Modeling}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {74} -} - -@incollection{grisafiAtomicscaleRepresentationStatistical2019, - author = {Grisafi, Andrea and Wilkins, David M. and Willatt, Michael J. and Ceriotti, Michele}, - booktitle = {Machine {{Learning}} in {{Chemistry}}: {{Data}}-{{Driven Algorithms}}, {{Learning Systems}}, and {{Predictions}}}, - date = {2019}, - pages = {1--21}, - publisher = {{ACS Publications}}, - title = {Atomic-Scale Representation and Statistical Learning of Tensorial Properties} -} - -@article{guAdaptiveIronbasedMagnetic2021, - abstract = {With unique physicochemical properties and biological effects, magnetic nanomaterials (MNMs) play a crucial role in the biomedical field. In particular, magnetic iron oxide nanoparticles (MIONPs) are approved by the United States Food and Drug Administration (FDA) for clinical applications at present due to their low toxicity, biocompatibility, and biodegradability. Despite the unarguable effectiveness, massive space for improving such materials' performance still needs to be filled. Recently, many efforts have been devoted to improving the preparation methods based on the materials' biosafety. Besides, researchers have successfully regulated the performance of magnetic nanoparticles (MNPs) by changing their sizes, morphologies, compositions; or by aggregating as-synthesized MNPs in an orderly arrangement to meet various clinical requirements. The rise of cloud computing and artificial intelligence techniques provides novel ways for fast material characterization, automated data analysis, and mechanism demonstration. In this review, we summarized the studies that focused on the preparation routes and performance regulations of high-quality MNPs, and their special properties applied in biomedical detection, diagnosis, and treatment. At the same time, the future development of MNMs was also discussed.}, - annotation = {WOS:000663236300002}, - author = {Gu, Ning and Zhang, Zuoheng and Li, Yan}, - date = {2021}, - doi = {10.1007/s12274-021-3546-1}, - issn = {1998-0124}, - journaltitle = {Nano Research}, - langid = {english}, - location = {{Beijing}}, - publisher = {{Tsinghua Univ Press}}, - shortjournal = {Nano Res.}, - title = {Adaptive Iron-Based Magnetic Nanomaterials of High Performance for Biomedical Applications}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06} -} - -@thesis{hanDeepLearningLargescale2018, - author = {Han, Jiequn}, - date = {2018}, - institution = {{Princeton University}}, - title = {Deep {{Learning}} for {{Large}}-Scale {{Molecular Dynamics}} and {{High}}-Dimensional {{Partial Differential Equations}}}, - type = {PhD Thesis} -} - -@article{hartMachineLearningAlloys2021, - abstract = {Alloy modelling has a history of machine-learning-like approaches, preceding the tide of data-science-inspired work. The dawn of computational databases has made the integration of analysis, prediction and discovery the key theme in accelerated alloy research. Advances in machine-learning methods and enhanced data generation have created a fertile ground for computational materials science. Pairing machine learning and alloys has proven to be particularly instrumental in pushing progress in a wide variety of materials, including metallic glasses, high-entropy alloys, shape-memory alloys, magnets, superalloys, catalysts and structural materials. This Review examines the present state of machine-learning-driven alloy research, discusses the approaches and applications in the field and summarizes theoretical predictions and experimental validations. We foresee that the partnership between machine learning and alloys will lead to the design of new and improved systems. Machine learning is enabling a metallurgical renaissance. This Review discusses recent progress in representations, descriptors and interatomic potentials, overviewing metallic glasses, high-entropy alloys, superalloys and shape-memory alloys, magnets and catalysts, and the prediction of mechanical and thermal properties.}, - annotation = {WOS:000675035000001}, - author = {Hart, Gus L. W. and Mueller, Tim and Toher, Cormac and Curtarolo, Stefano}, - date = {2021}, - doi = {10.1038/s41578-021-00340-w}, - issn = {2058-8437}, - journaltitle = {Nature Reviews Materials}, - langid = {english}, - location = {{Berlin}}, - publisher = {{Nature Research}}, - shortjournal = {Nat. Rev. Mater.}, - title = {Machine Learning for Alloys}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06} -} - -@incollection{heryadiCharacterizingPerformanceImprovement2019, - author = {Heryadi, Dodi and Hampton, Scott}, - booktitle = {Proceedings of the {{Practice}} and {{Experience}} in {{Advanced Research Computing}} on {{Rise}} of the {{Machines}} (Learning)}, - date = {2019}, - pages = {1--5}, - title = {Characterizing {{Performance Improvement}} of {{GPUs}}} -} - -@article{karniadakisPhysicsinformedMachineLearning2021, - abstract = {The rapidly developing field of physics-informed learning integrates data and mathematical models seamlessly, enabling accurate inference of realistic and high-dimensional multiphysics problems. This Review discusses the methodology and provides diverse examples and an outlook for further developments. Despite great progress in simulating multiphysics problems using the numerical discretization of partial differential equations (PDEs), one still cannot seamlessly incorporate noisy data into existing algorithms, mesh generation remains complex, and high-dimensional problems governed by parameterized PDEs cannot be tackled. Moreover, solving inverse problems with hidden physics is often prohibitively expensive and requires different formulations and elaborate computer codes. Machine learning has emerged as a promising alternative, but training deep neural networks requires big data, not always available for scientific problems. Instead, such networks can be trained from additional information obtained by enforcing the physical laws (for example, at random points in the continuous space-time domain). Such physics-informed learning integrates (noisy) data and mathematical models, and implements them through neural networks or other kernel-based regression networks. Moreover, it may be possible to design specialized network architectures that automatically satisfy some of the physical invariants for better accuracy, faster training and improved generalization. Here, we review some of the prevailing trends in embedding physics into machine learning, present some of the current capabilities and limitations and discuss diverse applications of physics-informed learning both for forward and inverse problems, including discovering hidden physics and tackling high-dimensional problems.}, - annotation = {WOS:000653612800001}, - author = {Karniadakis, George Em and Kevrekidis, Ioannis G. and Lu, Lu and Perdikaris, Paris and Wang, Sifan and Yang, Liu}, - date = {2021-06}, - doi = {10.1038/s42254-021-00314-5}, - journaltitle = {Nature Reviews Physics}, - langid = {english}, - location = {{London}}, - number = {6}, - pages = {422--440}, - publisher = {{Springernature}}, - shortjournal = {Nat. Rev. Phys.}, - title = {Physics-Informed Machine Learning}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {3} -} - -@article{Kocer2021, - abstract = {In the past two decades, machine learning potentials (MLP) have reached a level of maturity that now enables applications to large-scale atomistic simulations of a wide range of systems in chemistry, physics and materials science. Different machine learning algorithms have been used with great success in the construction of these MLPs. In this review, we discuss an important group of MLPs relying on artificial neural networks to establish a mapping from the atomic structure to the potential energy. In spite of this common feature, there are important conceptual differences, which concern the dimensionality of the systems, the inclusion of long-range electrostatic interactions, global phenomena like non-local charge transfer, and the type of descriptor used to represent the atomic structure, which can either be predefined or learnable. A concise overview is given along with a discussion of the open challenges in the field.}, - author = {Kocer, Emir and Ko, TW Tsz Wai and Behler, Jörg and Behler, J}, - date = {2021-07}, - journaltitle = {arxiv.org}, - title = {Neural Network Potentials: {{A}} Concise Overview of Methods}, - url = {https://arxiv.org/abs/2107.03727 http://arxiv.org/abs/2107.03727} -} - -@thesis{koFirstPrinciplesStudyStructural2019, - author = {Ko, Hsin-Yu}, - date = {2019}, - institution = {{Princeton University}}, - title = {First-{{Principles Study}} on the {{Structural}} and {{Thermal Properties}} of {{Molecular Crystals}} and {{Liquids}}}, - type = {PhD Thesis} -} - -@incollection{komeijiFMOInterfacedMolecular2021, - abstract = {Three ways to combine FMO and MD are described: FMO-MD, FMO-QM/MM-MD, and MM-MD/FMO. FMO-MD is an ab initio MD in which force is updated on-the-fly by FMO. FMO-QM/MM-MD is a QM/MM-MD method in which the QM part is calculated by FMO. MM-MD/FMO is a simulation protocol in which FMO calculation is performed for molecular configurations generated by MM-MD. The methodology and application of these methods are described and compared.}, - author = {Komeiji, Yuto and Ishikawa, Takeshi}, - booktitle = {Recent {{Advances}} of the {{Fragment Molecular Orbital Method}}}, - date = {2021}, - pages = {373--389}, - publisher = {{Springer}}, - title = {{{FMO Interfaced}} with {{Molecular Dynamics Simulation}}}, - url = {https://link.springer.com/chapter/10.1007/978-981-15-9235-5_19} -} - -@thesis{lefdalsnesClassicalMolecularDynamics2019, - author = {Lefdalsnes, Andreas Godø}, - date = {2019}, - title = {Classical {{Molecular Dynamics}} Using {{Neural Network Representations}} of {{Potential Energy Surfaces}}}, - type = {Master's Thesis} -} - -@article{leModelingElectrifiedMetal2021, - abstract = {The structure and potential distribution of electric double layers (EDLs) are of close relevance to the performance of electrode materials. In the past years, despite tremendous efforts devoted to this topic, an atomistic picture of the EDL is still lacking, let alone understanding on how the EDL structure is related to the dielectric property of interface water. In this article, we briefly review the recent progress in modeling electrified metal/water interfaces using ab initio molecular dynamics (AIMD). The ab initio methods for EDL modeling is firstly summarized, and then we discuss the structures of interface water on metal electrodes at different potential conditions. Moreover, we illustrate the potential-dependent behavior of chemisorbed water on Pt(111) surface and its relationship with the peak of the differential Helmholtz capacitance observed by experiment. At last, we give some perspective for future development in ab initio modeling of electrochemical interfaces.}, - author = {Le, Jia-Bo and Cheng, Jun}, - date = {2021-06}, - doi = {10/ghtqnk}, - issn = {24519103}, - journaltitle = {Current Opinion in Electrochemistry}, - langid = {english}, - pages = {100693}, - shortjournal = {Current Opinion in Electrochemistry}, - shorttitle = {Modeling Electrified Metal/Water Interfaces from Ab Initio Molecular Dynamics}, - title = {Modeling Electrified Metal/Water Interfaces from Ab Initio Molecular Dynamics: {{Structure}} and {{Helmholtz}} Capacitance}, - url = {https://linkinghub.elsevier.com/retrieve/pii/S2451910321000077}, - urldate = {2021-08-11}, - volume = {27} -} - -@thesis{liMolecularDynamicsStudy2021, - abstract = {Although electrostatics interactions in uids have been studied for many decades, new results in this field are still challenging our classical understanding of electrolytes. The combination of electrostatics and self-assembly yields many interesting yet challenging problems that are of fundamental scienti -c interest and show promise for industrial applications. In this dissertation, I introduce our work on the topic of charged nanomaterials in aqueous salt solutions and how electrostatics play a role in different systems. The work summarized here is an attempt to develop methods to correctly model nanoscale charged systems in both low and high salt environments, and tackle the problem of simulating large systems with MD simulations by using coarse-graining techniques. First, we study nanoparticles immersed in concentrated monovalent salt ({$>$}0.5 mol/L1) using multi-scale molecular dynamics (MD) simulations involving atomic resolution and coarse-grained representations with implicit solvent. We attaind a surprising attractive to repulsive and then attractive re-entrant behavior as a function of salt concentration that cannot be predicted by previous theories and propose a rational explanation. Next, we explore the interaction of cylindrical interfaces in NaCl solutions to -nd the screening length of charged cylinders and compare them with the prediction of the Poisson-Boltzmann (PB) equation. We also -nd a depletion attraction between cylinders at high monovalent salt concentrations. We compare the results of MD simulations to mean-field theories as well as liquid state theory that incorporates ion correlations, and we show that the short-range ion correlations significantly impact the interactions between cylinders in concentrated monovalent salt solutions. Finally, we look into the complex biological system of bacterial microcompartment (MCP) assembly. Using all-atom (AA, explicit water, and ion) and coarse-grained (CG, implicit ion) MD simulations, combined with thermodynamics analysis, we find that electrostatic interactions (hydrogen bonds and charge distributions) play an important role in the self-assembly of native propanediol utilization (Pdu) MCPs. Combining AA and CG MD simulations, we predict various polyhedral and extended assembly shapes, and we predict what kinds of mutations lead to the success or failure of MCP assembly. The simulation and theoretical predictions match with the experimental observation of our collaborators and with published experiments.}, - author = {Li, Yaohua}, - date = {2021}, - institution = {{Northwestern University}}, - shorttitle = {Molecular {{Dynamics Study}} of {{Charged Nanomaterials}}}, - title = {Molecular {{Dynamics Study}} of {{Charged Nanomaterials}}: {{Electrostatics}} and {{Self}}-{{Assembly}}}, - type = {PhD Thesis} -} - -@article{louieDiscoveringUnderstandingMaterials2021, - abstract = {Materials modelling and design using computational quantum and classical approaches is by now well established as an essential pillar in condensed matter physics, chemistry and materials science research, in addition to experiments and analytical theories. The past few decades have witnessed tremendous advances in methodology development and applications to understand and predict the ground-state, excited-state and dynamical properties of materials, ranging from molecules to nanoscopic/mesoscopic materials to bulk and reduced-dimensional systems. This issue of Nature Materials presents four in-depth Review Articles on the field. This Perspective aims to give a brief overview of the progress, as well as provide some comments on future challenges and opportunities. We envision that increasingly powerful and versatile computational approaches, coupled with new conceptual understandings and the growth of techniques such as machine learning, will play a guiding role in the future search and discovery of materials for science and technology. This Perspective provides an overview of the different approaches used to understand the behaviour of materials at different length scales and timescales through computation, and outlines future challenges in the description of complex systems or ultrafast non-equilibrium behaviour.}, - annotation = {WOS:000655912800009}, - author = {Louie, Steven G. and Chan, Yang-Hao and da Jornada, Felipe H. and Li, Zhenglu and Qiu, Diana Y.}, - date = {2021-06}, - doi = {10.1038/s41563-021-01015-1}, - issn = {1476-1122}, - journaltitle = {Nature Materials}, - langid = {english}, - location = {{Berlin}}, - number = {6}, - options = {useprefix=true}, - pages = {728--735}, - publisher = {{Nature Research}}, - shortjournal = {Nat. Mater.}, - title = {Discovering and Understanding Materials through Computation}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {20} -} - -@article{luFutureDirectionsChemical2021, - author = {Lu, Yuyuan and Deng, Geng and Shuai, Zhigang}, - date = {2021}, - doi = {10.1515/pac-2020-1006}, - journaltitle = {Pure and Applied Chemistry}, - publisher = {{De Gruyter}}, - title = {Future Directions of Chemical Theory and Computation} -} - -@thesis{luIntegratingMachineLearning2020, - author = {Lu, Jianing}, - date = {2020}, - institution = {{New York University}}, - title = {Integrating {{Machine Learning}} into {{Protein}}-{{Ligand Scoring Function Development}}}, - type = {PhD Thesis} -} - -@article{martinaDevelopmentMachineLearning, - abstract = {Recent experimental research showed that nucleotides, under favorable conditions of temperature and concentration, can self-assemble into liquid crystals. The mechanism involves the stacking of nucleotides into columnar aggregates. It has been proposed that this ordered structure can favor the polymerization of long nucleotide chains, which is a fundamental step toward the so called “RNA world”. In this thesis, starting from ab initio molecular dynamics simulations, at the density functional theory level, an all-atom potential for nucleotides in water, based on an implicit neural network representation, has been developed. Its stability and accuracy have been tested and its predictions on simple model systems have been compared with data generated both ab initio and using currently available empirical force field for nucleic acids.}, - author = {Martina, Riccardo}, - langid = {english}, - pages = {57}, - title = {Development of a Machine Learning Potential for Nucleotides in Water} -} - -@article{meuwlyMachineLearningChemical2021, - abstract = {Machine learning (ML) techniques applied to chemical reactions have a long history. The present contribution discusses applications ranging from small molecule reaction dynamics to computational platforms for reaction planning. ML-based techniques can be particularly relevant for problems involving both computation and experiments. For one, Bayesian inference is a powerful approach to develop models consistent with knowledge from experiments. Second, ML-based methods can also be used to handle problems that are formally intractable using conventional approaches, such as exhaustive characterization of state-to-state information in reactive collisions. Finally, the explicit simulation of reactive networks as they occur in combustion has become possible using machine-learned neural network potentials. This review provides an overview of the questions that can and have been addressed using machine learning techniques, and an outlook discusses challenges in this diverse and stimulating field. It is concluded that ML applied to chemistry problems as practiced and conceived today has the potential to transform the way with which the field approaches problems involving chemical reactions, in both research and academic teaching.}, - author = {Meuwly, Markus}, - date = {2021}, - doi = {10.1021/acs.chemrev.1c00033}, - journaltitle = {Chemical Reviews}, - publisher = {{ACS Publications}}, - title = {Machine {{Learning}} for {{Chemical Reactions}}}, - url = {https://pubs.acs.org/doi/abs/10.1021/acs.chemrev.1c00033} -} - -@article{mikschStrategiesConstructionMachinelearning2021, - abstract = {Recent advances in machine-learning interatomic potentials have enabled the efficient modeling of complex atomistic systems with an accuracy that is comparable to that of conventional quantum-mechanics based methods. At the same time, the construction of new machine-learning potentials can seem a daunting task, as it involves data-science techniques that are not yet common in chemistry and materials science. Here, we provide a tutorial-style overview of strategies and best practices for the construction of artificial neural network (ANN) potentials. We illustrate the most important aspects of (a) data collection, (b) model selection, (c) training and validation, and (d) testing and refinement of ANN potentials on the basis of practical examples. Current research in the areas of active learning and delta learning are also discussed in the context of ANN potentials. This tutorial review aims at equipping computational chemists and materials scientists with the required background knowledge for ANN potential construction and application, with the intention to accelerate the adoption of the method, so that it can facilitate exciting research that would otherwise be challenging with conventional strategies.}, - annotation = {WOS:000674925500001}, - author = {Miksch, April M. and Morawietz, Tobias and Kaestner, Johannes and Urban, Alexander and Artrith, Nongnuch}, - date = {2021-09}, - doi = {10.1088/2632-2153/abfd96}, - journaltitle = {Machine Learning-Science and Technology}, - langid = {english}, - location = {{Bristol}}, - number = {3}, - pages = {031001}, - publisher = {{Iop Publishing Ltd}}, - shortjournal = {Mach. Learn.-Sci. Technol.}, - title = {Strategies for the Construction of Machine-Learning Potentials for Accurate and Efficient Atomic-Scale Simulations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {2} -} - -@article{moqadamMembraneModelsMolecular2021, - abstract = {Peripheral membrane proteins (PMPs) bind temporarily to the surface of biological membranes. They also exist in a soluble form and their tertiary structure is often known. Yet, their membrane-bound form and their interfacial-binding site with membrane lipids remain difficult to observe directly. Their binding and unbinding mechanism, the conformational changes of the PMPs and their influence on the membrane structure are notoriously challenging to study experimentally. Molecular dynamics simulations are particularly useful to fill some knowledge-gaps and provide hypothesis that can be experimentally challenged to further our understanding of PMP-membrane recognition. Because of the time-scales of PMP-membrane binding events and the computational costs associated with molecular dynamics simulations, membrane models at different levels of resolution are used and often combined in multiscale simulation strategies. We here review membrane models belonging to three classes: atomistic, coarse-grained and implicit. Differences between models are rooted in the underlying theories and the reference data they are parameterized against. The choice of membrane model should therefore not only be guided by its computational efficiency. The range of applications of each model is discussed and illustrated using examples from the literature. [GRAPHICS] .}, - annotation = {WOS:000669104100001}, - author = {Moqadam, Mahmoud and Tubiana, Thibault and Moutoussamy, Emmanuel E. and Reuter, Nathalie}, - date = {2021-01-01}, - doi = {10.1080/23746149.2021.1932589}, - issn = {2374-6149}, - journaltitle = {Advances in Physics-X}, - langid = {english}, - location = {{Abingdon}}, - number = {1}, - pages = {1932589}, - publisher = {{Taylor \& Francis Ltd}}, - shortjournal = {Adv. Phys.-X}, - title = {Membrane Models for Molecular Simulations of Peripheral Membrane Proteins}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {6} -} - -@article{morawietzMachineLearningacceleratedQuantum2021, - abstract = {Atomistic simulations have become an invaluable tool for industrial applications ranging from the optimization of protein-ligand interactions for drug discovery to the design of new materials for energy applications. Here we review recent advances in the use of machine learning (ML) methods for accelerated simulations based on a quantum mechanical (QM) description of the system. We show how recent progress in ML methods has dramatically extended the applicability range of conventional QM-based simulations, allowing to calculate industrially relevant properties with enhanced accuracy, at reduced computational cost, and for length and time scales that would have otherwise not been accessible. We illustrate the benefits of ML-accelerated atomistic simulations for industrial R\&D processes by showcasing relevant applications from two very different areas, drug discovery (pharmaceuticals) and energy materials. Writing from the perspective of both a molecular and a materials modeling scientist, this review aims to provide a unified picture of the impact of ML-accelerated atomistic simulations on the pharmaceutical, chemical, and materials industries and gives an outlook on the exciting opportunities that could emerge in the future.}, - annotation = {WOS:000577932000001}, - author = {Morawietz, Tobias and Artrith, Nongnuch}, - date = {2021-04}, - doi = {10.1007/s10822-020-00346-6}, - issn = {0920-654X}, - journaltitle = {Journal of Computer-Aided Molecular Design}, - langid = {english}, - location = {{Dordrecht}}, - number = {4}, - pages = {557--586}, - publisher = {{Springer}}, - shortjournal = {J. Comput.-Aided Mol. Des.}, - title = {Machine Learning-Accelerated Quantum Mechanics-Based Atomistic Simulations for Industrial Applications}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {35} -} - -@article{musilPhysicsinspiredStructuralRepresentations2021, - abstract = {The first step in the construction of a regression model or a data-driven analysis, aiming to predict or elucidate the relationship between the atomic-scale structure of matter and its properties, involves transforming the Cartesian coordinates of the atoms into a suitable representation. The development of atomic-scale representations has played, and continues to play, a central role in the success of machine-learning methods for chemistry and materials science. This review summarizes the current understanding of the nature and characteristics of the most commonly used structural and chemical descriptions of atomistic structures, highlighting the deep underlying connections between different frameworks and the ideas that lead to computationally efficient and universally applicable models. It emphasizes the link between properties, structures, their physical chemistry, and their mathematical description, provides examples of recent applications to a diverse set of chemical and materials science problems, and outlines the open questions and the most promising research directions in the field.}, - author = {Musil, Felix and Grisafi, Andrea and Bartók, Albert P. and Ortner, Christoph and Csányi, Gábor and Ceriotti, Michele}, - date = {2021}, - doi = {10.1021/acs.chemrev.1c00021}, - journaltitle = {Chemical Reviews}, - publisher = {{ACS Publications}}, - title = {Physics-Inspired Structural Representations for Molecules and Materials}, - url = {https://pubs.acs.org/doi/abs/10.1021/acs.chemrev.1c00021} -} - -@thesis{pulighedduComputationalPredictionsThermal2020, - author = {Puligheddu, Marcello}, - date = {2020}, - institution = {{The University of Chicago}}, - title = {Computational {{Predictions}} of the {{Thermal Conductivity}} of {{Solids}} and {{Liquids}}}, - type = {PhD Thesis} -} - -@thesis{roccaRANDOMPHASEAPPROXIMATION2020, - abstract = {Despite the high computational cost the adiabatic connection fluctuation dissipation theorem (ACFDT) represents a promising approach to improve the description of the electronic correlation within density functional theory. The simplest approximation that can be applied in the context of the ACFDT is the random phase approximation (RPA). First, we show how the RPA can be improved by introducing a kernel containing an approximate electron-hole exchange term that leads to two different beyond-RPA methods. Then, we show that the RPA and beyond-RPA approaches can be efficiently computed within a plane-wave basis set implementation by using dielectric eigenpotentials as a compact auxiliary basis set and the Lanczos algorithm. A series of applications to molecules and solids are presented to demonstrate the efficiency and accuracy of these approximations. Importantly, it will be shown that the highly accurate beyond-RPA methods can be scaled to treat molecular systems with one hundred electrons requiring a basis set with hundreds of thousands of plane-waves. Finally, it is shown how the sophisticated and computationally expensive ACFDT methods can be used to compute finite-temperature properties of realistic materials (adsorption enthalpies of molecules in zeolites) by coupling molecular-dynamics simulations with machine learning algorithms.}, - author = {Rocca, Dario}, - date = {2020}, - institution = {{Université de Lorraine}}, - shorttitle = {{{RANDOM PHASE APPROXIMATION AND BEYOND}}}, - title = {{{RANDOM PHASE APPROXIMATION AND BEYOND}}: {{FROM THEORY TO REALISTIC MATERIALS}}}, - type = {PhD Thesis} -} - -@article{rousseauTheoreticalInsightsSurface2020, - abstract = {Redox-active oxides find use in many applications, including catalysts, photovoltaic devices, self-cleaning glasses, chemical sensors and electronic components. Their utility derives from their unique ability to access multiple metal-charge states within a finite energy window. However, this property also confounds our ability to study reducible oxides, because it leads to structural, compositional and electronic complexities that elude simplistic models of materials structure and function. Oxygen vacancies play a critical role in shaping the functional properties of such oxides; most notably, they lead to mobile-charge imbalances that impact surface processes at substantial distances from the originating defect. Atomistic simulations are inherently equipped to illuminate these phenomena at a fundamental level; however, reducible oxides pose great challenges, owing to the high level of electron correlation needed to correctly describe them. Understanding how defects form, couple, propagate, agglomerate or repel each other and influence the surface properties of reducible oxides is only now coming into the grasp of modern theory and simulation capabilities. This knowledge is also key to discovering and controlling emergent materials properties with tunable multifunctionalities at the nanometre scale and beyond. Reducible oxides are tunable, multifunctional materials used in many applications, particularly in catalysis; their attractive properties arise from their interacting charge carriers, complex electronic structure and propensity to form mobile defects. This Review surveys theoretical methods to model and understand reducible oxides, using TiO2 as a prototypical example.}, - annotation = {WOS:000535869400002}, - author = {Rousseau, Roger and Glezakou, Vassiliki-Alexandra and Selloni, Annabella}, - date = {2020-06}, - doi = {10.1038/s41578-020-0198-9}, - issn = {2058-8437}, - journaltitle = {Nature Reviews Materials}, - langid = {english}, - location = {{London}}, - number = {6}, - pages = {460--475}, - publisher = {{Nature Publishing Group}}, - shortjournal = {Nat. Rev. Mater.}, - title = {Theoretical Insights into the Surface Physics and Chemistry of Redox-Active Oxides}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/4}, - urldate = {2021-08-06}, - volume = {5} -} - -@incollection{saucedaConstructionMachineLearned2020, - abstract = {Highly accurate force fields are a mandatory requirement to generate predictive simulations. Here we present the path for the construction of machine learned molecular force fields by discussing the hierarchical pathway from generating the dataset of reference calculations to the construction of the machine learning model, and the validation of the physics generated by the model. We will use the the symmetrized gradient-domain machine learning (sGDML) framework due to its ability to reconstruct complex high-dimensional potential-energy surfaces (PES) with high precision even when using just a few hundreds of molecular conformations for training. The data efficiency of the sGDML model allows using reference atomic forces computed with high-level wavefunction-based approaches, such as the gold standard coupled cluster method with single, double, and perturbative triple excitations (CCSD(T)). We demonstrate that the flexible nature of the sGDML framework captures local and non-local electronic interactions (e.g. H-bonding, lone pairs, steric repulsion, changes in hybridization states (e.g. sp2 sp3), interactions, and proton transfer) without imposing any restriction on the nature of interatomic potentials. The analysis of sGDML models trained for different molecular structures at different levels of theory (e.g. density functional theory and CCSD(T)) provides empirical evidence that a higher level of theory generates a smoother PES. Additionally, a careful analysis of molecular dynamics simulations yields new qualitative insights into dynamics and vibrational spectroscopy of small molecules close to spectroscopic accuracy.}, - author = {Sauceda, Huziel E. and Chmiela, Stefan and Poltavsky, Igor and Müller, Klaus-Robert and Tkatchenko, Alexandre}, - booktitle = {Machine {{Learning Meets Quantum Physics}}}, - date = {2020}, - pages = {277--307}, - publisher = {{Springer}}, - shorttitle = {Construction of Machine Learned Force Fields with Quantum Chemical Accuracy}, - title = {Construction of Machine Learned Force Fields with Quantum Chemical Accuracy: {{Applications}} and Chemical Insights} -} - -@article{shaiduInteratomicPotentialLiC2020, - author = {Shaidu, Yusuf}, - date = {2020}, - publisher = {{SISSA}}, - title = {Interatomic {{Potential}} for {{Li}}-{{C Systems}} from {{Cluster Expansion}} to {{Artificial Neural Network Techniques}}} -} - -@article{shaoModellingBulkElectrolytes2021, - abstract = {Batteries and supercapacitors are electrochemical energy storage systems which involve multiple time-scales and length-scales. In terms of the electrolyte which serves as the ionic conductor, a molecular-level understanding of the corresponding transport phenomena, electrochemical (thermal) stability and interfacial properties is crucial for optimizing the device performance and achieving safety requirements. To this end, atomistic machine learning is a promising technology for bridging microscopic models and macroscopic phenomena. Here, we provide a timely snapshot of recent advances in this area. This includes technical considerations that are particularly relevant for modelling electrolytes as well as specific examples of both bulk electrolytes and associated interfaces. A perspective on methodological challenges and new applications is also discussed.}, - annotation = {WOS:000604327200001}, - author = {Shao, Yunqi and Knijff, Lisanne and Dietrich, Florian M. and Hermansson, Kersti and Zhang, Chao}, - date = {2021-04}, - doi = {10.1002/batt.202000262}, - journaltitle = {Batteries \& Supercaps}, - langid = {english}, - location = {{Weinheim}}, - number = {4}, - pages = {585--595}, - publisher = {{Wiley-V C H Verlag Gmbh}}, - shortjournal = {Batteries Supercaps}, - title = {Modelling {{Bulk Electrolytes}} and {{Electrolyte Interfaces}} with {{Atomistic Machine Learning}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {4} -} - -@thesis{srivastavaNeuralNetworkPrediction2020, - abstract = {The trend of open material data and automation in the past decade offers a unique opportunity for data-driven design of novel materials for various applications as well as fundamental scientific understanding, but it also poses a challenge for conventional machine learning approaches based on structure features. In this thesis, I develop a class of deep learning methods that solve various types of learning problems for solid materials, and demonstrate its application to both accelerate material design and understand scientific knowledge. First, I present a neural network architecture to learn the representations of an arbitrary solid material, which encodes several fundamental symmetries for solid materials as inductive biases. Then, I extend the approach to explore four different learning problems: 1) supervised learning to predict material properties from structures; 2) visualization to understand structure-property relations; 3) unsupervised learning to understand atomic scale dynamics from time series trajectories; 4) active learning to explore an unknown material space. In each learning problem, I demonstrate the performance of the approach compared with previous approaches, and apply it to solve several realistic materials design problems and extract scientific insights from data.}, - author = {Srivastava, Gopal Narayan}, - date = {2020}, - institution = {{The Ohio State University}}, - title = {Neural Network for the Prediction of Force Differences between an Amino Acid in Solution and Vacuum}, - type = {PhD Thesis} -} - -@article{unkeMachineLearningForce2021, - abstract = {In recent years, the use of machine learning (ML) in computational chemistry has enabled numerous advances previously out of reach due to the computational complexity of traditional electronic-structure methods. One of the most promising applications is the construction of ML-based force fields (FFs), with the aim to narrow the gap between the accuracy of ab initio methods and the efficiency of classical FFs. The key idea is to learn the statistical relation between chemical structure and potential energy without relying on a preconceived notion of fixed chemical bonds or knowledge about the relevant interactions. Such universal ML approximations are in principle only limited by the quality and quantity of the reference data used to train them. This review gives an overview of applications of ML-FFs and the chemical insights that can be obtained from them. The core concepts underlying ML-FFs are described in detail, and a step-by-step guide for constructing and testing them from scratch is given. The text concludes with a discussion of the challenges that remain to be overcome by the next generation of ML-FFs.}, - author = {Unke, Oliver T. and Chmiela, Stefan and Sauceda, Huziel E. and Gastegger, Michael and Poltavsky, Igor and Schütt, Kristof T. and Tkatchenko, Alexandre and Müller, Klaus-Robert}, - date = {2021}, - doi = {10.1021/acs.chemrev.0c01111}, - journaltitle = {Chemical Reviews}, - publisher = {{ACS Publications}}, - title = {Machine Learning Force Fields}, - url = {https://pubs.acs.org/doi/abs/10.1021/acs.chemrev.0c01111} -} - -@article{vassilev-galindoChallengesMachineLearning2021, - abstract = {Dynamics of flexible molecules are often determined by an interplay between local chemical bond fluctuations and conformational changes driven by long-range electrostatics and van der Waals interactions. This interplay between interactions yields complex potential-energy surfaces (PESs) with multiple minima and transition paths between them. In this work, we assess the performance of the state-of-the-art Machine Learning (ML) models, namely, sGDML, SchNet, Gaussian Approximation Potentials/Smooth Overlap of Atomic Positions (GAPs/SOAPs), and Behler-Parrinello neural networks, for reproducing such PESs, while using limited amounts of reference data. As a benchmark, we use the cis to trans thermal relaxation in an azobenzene molecule, where at least three different transition mechanisms should be considered. Although GAP/SOAP, SchNet, and sGDML models can globally achieve a chemical accuracy of 1 kcal mol(-1) with fewer than 1000 training points, predictions greatly depend on the ML method used and on the local region of the PES being sampled. Within a given ML method, large differences can be found between predictions of close-to-equilibrium and transition regions, as well as for different transition mechanisms. We identify key challenges that the ML models face mainly due to the intrinsic limitations of commonly used atom-based descriptors. All in all, our results suggest switching from learning the entire PES within a single model to using multiple local models with optimized descriptors, training sets, and architectures for different parts of the complex PES.}, - annotation = {WOS:000630524000019}, - author = {Vassilev-Galindo, Valentin and Fonseca, Gregory and Poltavsky, Igor and Tkatchenko, Alexandre}, - date = {2021-03-07}, - doi = {10.1063/5.0038516}, - issn = {0021-9606}, - journaltitle = {Journal of Chemical Physics}, - langid = {english}, - location = {{Melville}}, - number = {9}, - pages = {094119}, - publisher = {{Amer Inst Physics}}, - shortjournal = {J. Chem. Phys.}, - title = {Challenges for Machine Learning Force Fields in Reproducing Potential Energy Surfaces of Flexible Molecules}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/2}, - urldate = {2021-08-06}, - volume = {154} -} - -@incollection{wangForceFieldDevelopment2019, - author = {Wang, Lee-Ping}, - booktitle = {Computational {{Approaches}} for {{Chemistry Under Extreme Conditions}}}, - date = {2019}, - pages = {127--159}, - publisher = {{Springer}}, - title = {Force {{Field Development}} and {{Nanoreactor Chemistry}}} -} - -@article{wangInvestigationsWaterOxide2021, - abstract = {Water/oxide interfaces are ubiquitous on earth and show significant influence on many chemical processes. For example, understanding water and solute adsorption as well as catalytic water splitting can help build better fuel cells and solar cells to overcome our looming energy crisis; the interaction between biomolecules and water/oxide interfaces is one hypothesis to explain the origin of life. However, knowledge in this area is still limited due to the difficulty of studying water/solid interfaces. As a result, research using increasingly sophisticated experimental techniques and computational simulations has been carried out in recent years. Although it is difficult for experimental techniques to provide detailed microscopic structural information, molecular dynamics (MD) simulations have satisfactory performance. In this review, we discuss classical and ab initio MD simulations of water/oxide interfaces. Generally, we are interested in the following questions: How do solid surfaces perturb interfacial water structure? How do interfacial water molecules and adsorbed solutes affect solid surfaces and how do interfacial environments affect solvent and solute behavior? Finally, we discuss progress in the application of neural network potential based MD simulations, which offer a promising future because this approach has already enabled ab initio level accuracy for very large systems and long trajectories. This article is categorized under: Theoretical and Physical Chemistry {$>$} Spectroscopy Molecular and Statistical Mechanics {$>$} Molecular Interactions Structure and Mechanism {$>$} Molecular Structures}, - annotation = {WOS:000657137500001}, - author = {Wang, Ruiyu and Klein, Michael L. and Carnevale, Vincenzo and Borguet, Eric}, - date = {2021}, - doi = {10.1002/wcms.1537}, - issn = {1759-0876}, - journaltitle = {Wiley Interdisciplinary Reviews-Computational Molecular Science}, - langid = {english}, - location = {{Hoboken}}, - pages = {e1537}, - publisher = {{Wiley}}, - shortjournal = {Wiley Interdiscip. Rev.-Comput. Mol. Sci.}, - title = {Investigations of Water/Oxide Interfaces by Molecular Dynamics Simulations}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06} -} - -@article{wangPhysicsGuidedDeepLearning2021, - abstract = {Modeling complex physical dynamics is a fundamental task in science and engineering. Traditional physics-based models are interpretable but rely on rigid assumptions. And the direct numerical approximation is usually computationally intensive, requiring significant computational resources and expertise. While deep learning (DL) provides novel alternatives for efficiently recognizing complex patterns and emulating nonlinear dynamics, it does not necessarily obey the governing laws of physical systems, nor do they generalize well across different systems. Thus, the study of physics-guided DL emerged and has gained great progress. It aims to take the best from both physics-based modeling and state-of-the-art DL models to better solve scientific problems. In this paper, we provide a structured overview of existing methodologies of integrating prior physical knowledge or physics-based modeling into DL and discuss the emerging opportunities.}, - archiveprefix = {arXiv}, - author = {Wang, Rui}, - date = {2021-08-01}, - eprint = {2107.01272}, - eprinttype = {arxiv}, - langid = {english}, - primaryclass = {cs}, - shorttitle = {Physics-{{Guided Deep Learning}} for {{Dynamical Systems}}}, - title = {Physics-{{Guided Deep Learning}} for {{Dynamical Systems}}: {{A}} Survey}, - url = {http://arxiv.org/abs/2107.01272}, - urldate = {2021-08-11} -} - -@article{Weinan2020, - abstract = {Machine learning is poised as a very powerful tool that can drastically improve our ability to carry out scientific research. However, many issues need to be addressed before this becomes a reality. This article focuses on one particular issue of broad interest: How can we integrate machine learning with physics-based modeling to develop new interpretable and truly reliable physical models? After introducing the general guidelines, we discuss the two most important issues for developing machine learning-based physical models: Imposing physical constraints and obtaining optimal datasets. We also provide a simple and intuitive explanation for the fundamental reasons behind the success of modern machine learning, as well as an introduction to the concurrent machine learning framework needed for integrating machine learning with physics-based modeling. Molecular dynamics and moment closure of kinetic equations are used as examples to illustrate the main issues discussed. We end with a general discussion on where this integration will lead us to, and where the new frontier will be after machine learning is successfully integrated into scientific modeling.}, - author = {Weinan, E and Han, Jiequn and Linfeng, Zhang}, - date = {2020-06}, - title = {Integrating Machine Learning with Physics-Based Modeling}, - url = {https://arxiv.org/abs/2006.02619 http://arxiv.org/abs/2006.02619 https://deepai.org/publication/integrating-machine-learning-with-physics-based-modeling} -} - -@article{weinanMachineLearningComputational2020, - abstract = {Neural network-based machine learning is capable of approximating functions in very high dimension with unprecedented efficiency and accuracy. This has opened up many exciting new possibilities, not just in traditional areas of artificial intelligence, but also in scientific computing and computational science. At the same time, machine learning has also acquired the reputation of being a set of "black box" type of tricks, without fundamental principles. This has been a real obstacle for making further progress in machine learning. In this article, we try to address the following two very important questions: (1) How machine learning has already impacted and will further impact computational mathematics, scientific computing and computational science? (2) How computational mathematics, particularly numerical analysis, can impact machine learning? We describe some of the most important progress that has been made on these issues. Our hope is to put things into a perspective that will help to integrate machine learning with computational mathematics.}, - annotation = {WOS:000592624200002}, - author = {Weinan, E.}, - date = {2020-11}, - doi = {10.4208/cicp.OA-2020-0185}, - issn = {1815-2406}, - journaltitle = {Communications in Computational Physics}, - langid = {english}, - location = {{Wanchai}}, - number = {5}, - pages = {1639--1670}, - publisher = {{Global Science Press}}, - shortjournal = {Commun. Comput. Phys.}, - title = {Machine {{Learning}} and {{Computational Mathematics}}}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {28} -} - -@article{westermayrMachineLearningElectronically2020, - abstract = {Electronically excited states of molecules are at the heart of photochemistry, photophysics, as well as photobiology and also play a role in material science. Their theoretical description requires highly accurate quantum chemical calculations, which are computationally expensive. In this review, we focus on not only how machine learning is employed to speed up such excited-state simulations but also how this branch of artificial intelligence can be used to advance this exciting research field in all its aspects. Discussed applications of machine learning for excited states include excited-state dynamics simulations, static calculations of absorption spectra, as well as many others. In order to put these studies into context, we discuss the promises and pitfalls of the involved machine learning techniques. Since the latter are mostly based on quantum chemistry calculations, we also provide a short introduction into excited-state electronic structure methods and approaches for nonadiabatic dynamics simulations and describe tricks and problems when using them in machine learning for excited states of molecules.}, - author = {Westermayr, Julia and Marquetand, Philipp}, - date = {2020}, - doi = {10.1021/acs.chemrev.0c00749}, - journaltitle = {Chemical Reviews}, - publisher = {{ACS Publications}}, - title = {Machine Learning for Electronically Excited States of Molecules}, - url = {https://pubs.acs.org/doi/abs/10.1021/acs.chemrev.0c00749} -} - -@article{Willard, - abstract = {There is a growing consensus that solutions to complex science and engineering problems require novel methodologies that are able to integrate traditional physics-based modeling approaches with state-of-the-art machine learning (ML) techniques. This paper provides a structured overview of such techniques. Applicationcentric objective areas for which these approaches have been applied are summarized, and then classes of methodologies used to construct physics-guided ML models and hybrid physics-ML frameworks are described. We then provide a taxonomy of these existing techniques, which uncovers knowledge gaps and potential crossovers of methods between disciplines that can serve as ideas for future research.}, - author = {Willard, J and Jia, X and Xu, S and Steinbach, M and Kumar, V}, - date = {2021}, - journaltitle = {arxiv.org}, - title = {Integrating Physics-Based Modeling with Machine Learning: {{A}} Survey}, - url = {https://arxiv.org/abs/2003.04919} -} - -@thesis{xieDeepLearningMethods2020, - abstract = {The trend of open material data and automation in the past decade offers a unique opportunity for data-driven design of novel materials for various applications as well as fundamental scientific understanding, but it also poses a challenge for conventional machine learning approaches based on structure features. In this thesis, I develop a class of deep learning methods that solve various types of learning problems for solid materials, and demonstrate its application to both accelerate material design and understand scientific knowledge. First, I present a neural network architecture to learn the representations of an arbitrary solid material, which encodes several fundamental symmetries for solid materials as inductive biases. Then, I extend the approach to explore four different learning problems: 1) supervised learning to predict material properties from structures; 2) visualization to understand structure-property relations; 3) unsupervised learning to understand atomic scale dynamics from time series trajectories; 4) active learning to explore an unknown material space. In each learning problem, I demonstrate the performance of the approach compared with previous approaches, and apply it to solve several realistic materials design problems and extract scientific insights from data.}, - author = {Xie, Tian}, - date = {2020}, - institution = {{Massachusetts Institute of Technology}}, - title = {Deep Learning Methods for the Design and Understanding of Solid Materials}, - type = {PhD Thesis} -} - -@article{xuPerspectiveComputationalReaction2021, - abstract = {Heterogeneous catalysis plays a significant role in the modern chemical industry. Towards the rational design of novel catalysts, understanding reactions over surfaces is the most essential aspect. Typical industrial catalytic processes such as syngas conversion and methane utilisation can generate a large reaction network comprising thousands of intermediates and reaction pairs. This complexity not only arises from the permutation of transformations between species but also from the extra reaction channels offered by distinct surface sites. Despite the success in investigating surface reactions at the atomic scale, the huge computational expense of ab initio methods hinders the exploration of such complicated reaction networks. With the proliferation of catalysis studies, machine learning as an emerging tool can take advantage of the accumulated reaction data to emulate the output of ab initio methods towards swift reaction prediction. Here, we briefly summarise the conventional workflow of reaction prediction, including reaction network generation, ab initio thermodynamics and microkinetic modelling. An overview of the frequently used regression models in machine learning is presented. As a promising alternative to full ab initio calculations, machine learning interatomic potentials are highlighted. Furthermore, we survey applications assisted by these methods for accelerating reaction prediction, exploring reaction networks, and computational catalyst design. Finally, we envisage future directions in computationally investigating reactions and implementing machine learning algorithms in heterogeneous catalysis.}, - annotation = {WOS:000648898600001}, - author = {Xu, Jiayan and Cao, Xiao-Ming and Hu, P.}, - date = {2021-05-21}, - doi = {10.1039/d1cp01349a}, - issn = {1463-9076}, - journaltitle = {Physical Chemistry Chemical Physics}, - langid = {english}, - location = {{Cambridge}}, - number = {19}, - pages = {11155--11179}, - publisher = {{Royal Soc Chemistry}}, - shortjournal = {Phys. Chem. Chem. Phys.}, - title = {Perspective on Computational Reaction Prediction Using Machine Learning Methods in Heterogeneous Catalysis}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/1}, - urldate = {2021-08-06}, - volume = {23} -} - -@article{yangRecentProgressMultiscale2021, - abstract = {Computational electrochemistry, an important branch of electrochemistry, has shown its advantages in studying electrode/electrolyte interfaces, such as the structures of electric double layers. However, modeling electrochemical systems is still a challenge, especially in interface electrochemistry, because not only solvation effects and ion distribution in electrolyte solutions should be considered, but also the treatment of the electrode potential and the response of electrolytes to applied potentials. Here, we review the latest development in the field of computational electrochemistry. We first introduce various energy models used in simulating electrolytes and electrodes at multiple scales. Then, to better explain and compare between different methods, we discuss the calculation methods of solution electrochemistry and interface electrochemistry in separate. At last, we introduce the methods to electrify the interfaces in various multiscale models. This review aims to help understand various levels of methods in simulations of different scenarios in electrochemistry, and summarizes a set of schemes covering multiple scales. This article is categorized under: Electronic Structure Theory ¿ Combined QM/MM Methods Molecular and Statistical Mechanics ¿ Molecular Dynamics and Monte-Carlo Methods Electronic Structure Theory ¿ Density Functional Theory.}, - author = {Yang, Xiao‐Hui and Zhuang, Yong‐Bin and Zhu, Jia‐Xin and Le, Jia‐Bo and Cheng, Jun}, - date = {2021-06}, - doi = {10.1002/wcms.1559}, - journaltitle = {WIREs Computational Molecular Science}, - publisher = {{Wiley}}, - title = {Recent Progress on Multiscale Modeling of Electrochemistry} -} - -@article{yeMachineLearningCoarseGrained2021, - author = {Ye, Huilin and Xian, Weikang and Li, Ying}, - date = {2021}, - doi = {10.1021/acsomega.0c05321}, - journaltitle = {ACS omega}, - number = {3}, - pages = {1758--1772}, - publisher = {{ACS Publications}}, - shorttitle = {Machine {{Learning}} of {{Coarse}}-{{Grained Models}} for {{Organic Molecules}} and {{Polymers}}}, - title = {Machine {{Learning}} of {{Coarse}}-{{Grained Models}} for {{Organic Molecules}} and {{Polymers}}: {{Progress}}, {{Opportunities}}, and {{Challenges}}}, - volume = {6} -} - -@article{Zhang2020c, - abstract = {In recent years, machine learning has emerged as a promising tool for dealing with the difficulty of representing high dimensional functions. This gives us an unprecedented opportunity to revisit theoretical foundations of various scientific fields, develop new schemes, improve existing methodologies, and solve problems that were too complicated for conventional approaches to address. In this dissertation, we identify a list of such problems in the context of multiscale molecular modeling and propose machine learning based strategies to boost simulations with ab initio accuracy to much larger scales than conventional approaches. We consider two representative challenges: 1) how to go from many-electron-ion to atomistic systems, for which the key has been a general and efficient representation of the potential energy surface generated by electronic structure models; 2) how to go from atomistic to coarse-grained systems, for which one is interested in the free energy of the coarse-grained variables as well as the associated dynamical behavior. Our strategies follow two seemingly obvious but non-trivial principles: 1) machine learning based models should respect important physical constraints like symmetry; 2) to build truly reliable models, efficient algorithms are needed to construct a minimal but truly representative training data set. We use these principles to construct the Deep Potential model for the potential energy surface, the Deep Potential Molecular dynamics (DeePMD) which is a new paradigm for performing ab initio molecular dynamics, a concurrent learning scheme (DP-GEN) for generating the data set on the fly, algorithms for constructing the Wannier centers (Deep Wanner) and for efficiently exploring the free energy landscape (Reinforced Dynamics), as well as a machine learning-based coarse grained molecular dynamics model (DeePCG), etc.Applications of these models and algorithms are presented for problems in chemistry, biology, and materials science. Finally, we present our efforts on developing related open-source software packages, which have now been widely used worldwide by experts and practitioners in the molecular simulation community.}, - author = {Zhang, L}, - date = {2020}, - title = {Machine Learning for Multi-Scale Molecular Modeling: Theories, Algorithms, and Applications}, - url = {https://search.proquest.com/openview/58ad7a1fdcc88005de25b81cd4cdc5d8/1?pq-origsite=gscholar&cbl=51922&diss=y} -} - -@article{zhangGlobalOptimizationChemical2021, - abstract = {Chemical clusters are relevant to many applications in catalysis, separations, materials, and energy sciences. Experimentally, the structure of clusters is difficult to determine, but it is very important in understanding their chemistry and properties. Computational methods can be used to examine cluster structure, however finding the most stable structure is not simple, particularly as the cluster size increases. Global optimization techniques have long been used to tackle the problem of the most stable structure, but such approaches would have to look for a global minimum, while sampling local minima over the whole potential energy surface as well. In this review, the state-of-the-art theory of global optimization theory is summarized. First, the definition, significance, relation to experiments, and a brief history of global optimization is presented. We then discuss, in more detail, three versatile global optimization methods: the basin hopping, the artificial bee colony algorithm, and the genetic algorithm. We close with some representative application examples of global optimization of clusters since 2016 and the challenges, open questions and opportunities in this field.}, - annotation = {WOS:000591191600001}, - author = {Zhang, Jun and Glezakou, Vassiliki-Alexandra}, - date = {2021-04-05}, - doi = {10.1002/qua.26553}, - issn = {0020-7608}, - journaltitle = {International Journal of Quantum Chemistry}, - langid = {english}, - location = {{Hoboken}}, - number = {7}, - pages = {e26553}, - publisher = {{Wiley}}, - shortjournal = {Int. J. Quantum Chem.}, - shorttitle = {Global Optimization of Chemical Cluster Structures}, - title = {Global Optimization of Chemical Cluster Structures: {{Methods}}, Applications, and Challenges}, - url = {https://www.webofscience.com/wos/alldb/summary/44a3404b-9775-4cfb-8e8f-3f1fab40830f-035e7f8d/date-descending/3}, - urldate = {2021-08-06}, - volume = {121} -} - -@thesis{zhangNoncontactUltrasound2019, - author = {Zhang, Xiang}, - date = {2019}, - institution = {{Massachusetts Institute of Technology}}, - title = {Non-Contact Ultrasound}, - type = {PhD Thesis} -} \ No newline at end of file diff --git a/source/papers/others.md b/source/papers/others.md deleted file mode 100644 index f6758f8b..00000000 --- a/source/papers/others.md +++ /dev/null @@ -1,8 +0,0 @@ ---- -title: Others -date: 2021-08-09 -update: 2021-08-09 ---- - - -{% publications_from_bib others.bib %} diff --git a/source/papers/reviews.md b/source/papers/reviews.md deleted file mode 100644 index 64e9afba..00000000 --- a/source/papers/reviews.md +++ /dev/null @@ -1,7 +0,0 @@ ---- -title: Reviews -date: 2021-08-09 -update: 2021-08-12 ---- - -{% publications_from_bib reviews.bib %}