Feature release 11 February 2026

This release includes the following changes:

• Update bundled Kokkos library to version 5.0.2 (Stan Moore, SNL) PR #4778, PR #4875
• New MBX package which is a wrapper around the MBX library for data driven many-body potentials (Henry Agnew, Francesco Paesani, UCSD, and Chris Knight, ANL) PR #4694 and PR #4871
• Add compute hbond/local to compute hydrogen bonds and output the list as local data (Axel Kohlmeyer, Temple U) PR #4867
• Significant updates to fix nvt/sllod, fix temp/deform, compute temp/deform and related commands for stable integration of the equations of motion in a way which is reversible and energy-conserving ( Stephen Sanderson, University of Queensland) PR #4781
• Add smooth termination of Coulomb integrals in the Streitz-Mintmire method (Roman Gröger, Czech Academy of Sciences) PR #4733
• Add support for use of fix shake with KOKKOS and minimization (Mitch Murphy) PR #4827
• Port minimization style fire to KOKKOS (Mitch Murphy) PR #4851
• Add support for ellipsoid atom style in KOKKOS (Lewis Russell, University of Strathclyde and Stan Moore,SNL) PR #4865
• Add a geometric integrator for overdamped rotational Brownian motion (Artur Straube, Zuse Institute Berlin) PR #4842
• Improved inter-bead communication for fix pimd/langevin (Yifan Li, Princeton) PR #4857
• Add support for wildcard atom types in fix bond/react and add an interface to type label handling that allows to infer bonded interaction type labels from the types of the constituent atom types (Jake Gissinger, Stevens IoT) PR #4822, PR #4879
• Significant enhancements to dump image (and dump movie) (Axel Kohlmeyer, Temple U) PR #4812 and others:
  • added an approximation to transparency (via "screen-door transparency" using a 16x16 Bayer matrix)
  • added support for visualizing data from fixes to: fix indent, fix reaxff/bonds, fix smd/wall_surface, fix wall/lj93, fix wall/lj126, fix wall/lj1043, fix wall/colloid, fix wall/gran, fix wall/harmonic, fix wall/harmonic/outside, fix wall/lepton, fix wall/morse, fix wall/reflect, fix wall/reflect/stochastic, and fix wall/table
  • added support for highlighting atoms involved in MC processes in fix atom/swap', fix bond/break', fix bond/create, fix bond/create/angle, fix mol/swap, and fix neighbor/swap (PR #4874)
  • several bugfixes
  • added a new fix graphics/objects command to place some graphics objects into the visualization including a progress bar and arrows
  • updates to body styles rounded/polygon and rounded/polyhedra to display faces plus code refactor (PR #4817)
  • support for ellipsoid particles through using triangle meshes (same as regions) also (PR #4817)
  • added a new fix graphics/arrows command to display per -atom or pre-chunk arrows (PR #4819)
  • added a new fix graphics/isosurface command to display isosurfaces of per-atom data and output them to STL files (PR #4823)
  • added a new fix graphics/labels command to display images or text inside the visualization (PR #4826 and PR #4848)
  • added a new fix graphics/lines command to display traces of averaged atom positions (aka trajectory lines) (PR #4876)
  • added a new fix graphics/periodic command to display periodic images of atoms and bonds (PR #4825)
  • added support for compute styles providing graphics objects to dump image and apply to new compute hbond/local (PR #4867)
  • added support for highlighting atoms in a reaction with fix bond/react (PR #4870)
  • added a "Advanced Graphics Howto" and visualization for plane regions (PR #4820)
  • added an option to dump image to create a vertical background gradient instead of a plain background (PR #4823)
  • consolidate generation of high-level graphics objects in an ImageObjects namespace and move to a separate file (PR #4820)
• Add support for compression with brotli and 7-zip through command line tools, add compression to write_data (Axel Kohlmeyer, Temple U) PR #4810
• Remove obsolete AMBER-to-LAMMPS python2 scripts and add Howto for using the external AMBER2LAMMPS software (Arun Sridhar, Heriot-Watt University) PR #4824
• Many small bug fixes, minor code improvements and refactoring, and updates for coding style, build system, documentation, and formatting (multiple authors): PR #4807, PR #4808, PR #4809, PR #4811, PR #4816, PR #4818, PR #4833, PR #4837, PR #4845, PR #4856, PR #4861, PR #4864, PR #4878, PR #4880, PR #4881

Backward compatibility notes:

• Since the Kokkos library version 5.0.0 and later dropped support for building with GNU make, the KOKKOS package in LAMMPS now also only supports building with CMake. The corresponding traditional GNU make build system files were removed or changed accordingly.
• As of version 5.0.0 Kokkos requires at least C++20
• The dump image and dump movie commands are now part of the new GRAPHICS package and thus the package must be installed to have access to them. It was added to the "basic" (and "most") preset and make yes-basic (and make yes-most), so this will affect few users
• The zcylinder keyword was removed from fix wall/gran since it is obsoleted by fix wall/gran/region which is more flexible.
• The amber2lammps.py, dump2trj99.py, and dump2trj.py scripts have been removed from the distribution. They were unmaintained, obsolete, and require Python 2.
• Since the new kick flag of fix nvt/sllod defaults to yes when using lab-frame velocity, scripts which previously applied the velocity profile manually would need to either stop doing that or set kick no.
• Previous scripts will also generate a warning about fix deform being applied at the end of the step, and suggest setting nevery = 0 in fix deform to correct the issue.
• With fix nvt/sllod now having compute temp/deform as a hard requirement, scripts which used some other velocity bias will need to adapt by setting that bias via the new temp argument of compute temp/deform.

---

Full Changelog: patch_10Dec2025...patch_11Feb2026

---

Information on the files attached to this release:

• lammps-src-11Feb2026.tar.gz - source code with translated manual in HTML and PDF format included
• lammps-linux-x86_64-11Feb2026.tar.gz - Fully static LAMMPS binaries for all Linux machines with x86_64 CPUs
• Manual_11Feb2026.pdf - LAMMPS manual as PDF
• lammps-gui-manual-v1.9.1.pdf - Manual for LAMMPS-GUI version v1.9.1 as PDF
• The following packages contain LAMMPS 11Feb2026 and LAMMPS-GUI v1.9.1 pre-compiled in different packaging formats
  • LAMMPS-Linux-x86_64-GUI-11Feb2026.tar.gz - Binaries with required libraries included using wrapper scripts. Compatible with Ubuntu 22.04LTS or later and equivalent Linux distributions
  • LAMMPS-Linux-x86_64-GUI-11Feb2026.flatpak - Linux binaries in flatpak bundle format and desktop integration
  • LAMMPS-macOS-multiarch-GUI-11Feb2026.dmg - DMG installer for macOS version 11 (Big Sur) or later. Includes support for Intel and Apple CPUs
  • LAMMPS-Win10-x64-GUI-11Feb2026.exe - Installer package for 64-bit x86 Windows 10 or later

If you are looking for a pre-compiled LAMMPS-GUI version that can load a custom compiled LAMMPS shared library (version 22 Jul 2025 update 2 or later) please go to the LAMMPS-GUI release page

Update 3 for Stable release 22 July 2025

This is the third update to the 22 July 2025 stable release that fixes bugs that have been reported since the second update and backported from the development version.

The following individual changes are included:

• Update list of installed header files and generated CMake configuration files
• Improvements and corrections for creating binary packages on Linux
• Backport INTERLAYER package bugfix
• Backport KOKKOS MEAM hybrid indexing bugfix
• Backport of GPU package corrections to exclusion handling and similar from upstream
• backport pair hybrid and bond hybrid extra value bugfix from upstream
• Bugfix for fix rigid/small with langevin option
• Bugfix for set image. Fixes #4834
• Bugfix for compiling SCAFACOS package
• Correctly accept kspace_modify splittol which was named eigtol in the source code
• avoid segfault when using fix shake during minimization without stats output
• Disallow use of INTEL package pair styles with MC package fixes that call Pair::compute()
• Add GitHub workflow for testing compilation with GNU make build system
• Reduce ccache sizes for GitHub runners to 3GB
• Backport CMake and packaging corrections from upstream
• Backport of LAMMPS-GUI fixes and selected improvements from upstream and label version as 1.7.3
• only use version dates on stable and release branch manual output, otherwise use Development or Maintenance
• use branch names when selecting the version of the online manual, add source snapshot download

---

Full Changelog: stable_22Jul2025_update2...stable_22Jul2025_update3

---

Information on the files attached to this release:

• lammps-src-22Jul2025_update3.tar.gz - LAMMPS source code with translated manual in HTML and PDF format included
• lammps-linux-x86_64-22Jul2025_update3.tar.gz - Fully static LAMMPS binaries for all Linux machines with x86_64 CPUs
• Manual_22July2025.pdf - PDF version of full manual for LAMMPS
• lammps-gui-v1.7.3-manual.pdf - Manual for LAMMPS-GUI only extracted from the LAMMPS manual above
• The following packages contain LAMMPS 22Jul2025_update3 and LAMMPS-GUI v1.7.3 pre-compiled in different packaging formats
  • LAMMPS-Linux-x86_64-GUI-22Jul2025_update3.tar.gz - Binaries with required libraries included using wrapper scripts. Compatible with Ubuntu 20.04LTS or later and equivalent Linux distributions
  • LAMMPS-Linux-x86_64-GUI-22Jul2025_update3.flatpak - Linux binaries in flatpak bundle format and desktop integration
  • LAMMPS-macOS-multiarch-GUI-22Jul2025_update3.dmg - DMG installer for macOS version 11 (Big Sur) or later. Includes support for Intel and Apple CPUs
  • LAMMPS-Win10-64bit-GUI-22Jul2025_update3.exe - Installer package for 64-bit x86 Windows 10 or later (may work on older Windows versions)

If you are looking for a pre-compiled LAMMPS-GUI version that can load a custom compiled LAMMPS shared library please go to the LAMMPS-GUI release page. Standalone LAMMPS-GUI versions prior to 1.8.4 are not compatible with this release, all later versions are.

Feature release 10 December 2025

This release includes the following changes:

• Update bundled Kokkos library to version 4.7.1 (Stan Moore, SNL) PR #4700
• Improve support for hybrid atom styles in KOKKOS package (Stan Moore, SNL) PR #4772
• Add new command write_molecule (Axel Kohlmeyer, Temple U) PR #4735
• Add comprehensive FFT testing for KSPACE/KOKKOS (Michael Lostica) PR #4728
• Add support for non-blocking FFT communication (Nick Hagerty, ORNL) PR #4789
• Add support for > 2B rigid bodies to fix rigid/small (Michael Lostica) PR #4755
• Add auto keyword for thermo output to have some more informative output about what is computed instead of the ID and a number (Jake Gissinger, Stevens IoT) PR #3764
• Add new pair style for van der Waals interactions at transition-metal dichalcogenide / metal interfaces (Wengen Ouyang, Wuhan University and Xiaohui Duan, Shandong University) PR #4780
• Add new fix style ttm/thermal (Bradly Baer and Greg Walker, Vanderbilt) PR #4689
• Add new fix style wall/harmonic/outside (Eddy Barraud) PR #3899
• Add new fix style align/self for self-propelled particles (Jeremy Fersula, Sorbonne) PR #4792
• Add new fix styles settorque/atom and addtorque/atom for manipulating per particle torque (Joel Clemmer, SNL) PR #4795
• Add KOKKOS support to pair style hybrid/scaled (Meg McCarthy, SNL) PR #4664
• Add KOKKOS support to compute temp/com (Stan Moore, SNL) PR #4768
• Add KOKKOS support for dihedral style fourier, improper style cvff and fix external (Stan Moore, SNL and Axel Kohlmeyer, Temple U) PR #4796
• Add more parallelization to pack/unpack for ML-IAP with KOKKOS (Chuck Witt, Harvard) PR #4744
• Add feature to output file with deleted molecules in JSON format to fix reaxff/species (Jake Gissinger, Stevens IoT) PR #4640
• Add feature to delete whole molecules in fix reaxff/species (Jake Gissinger, Stevens IoT) PR #4775
• Add feature to output file with reactions as molecules in JSON format to fix bond/react (Jake Gissinger, Stevens IoT) PR #4802
• Fix bond/react refactor and some argument adjustments (Jake Gissinger, Stevens IoT) PR #4750
• Multiple updates to fix pimd/langevin (Yifan Li and Kehan Cai, Princeton) PR #4722, PR #4756
• Fix srd bugfix for periodic boundaries and add unbiased thermostat (Victor Taenzel, University of Freiburg) PR #4590
• Provide access to constraint/restraint force in fix shake/rattle (Axel Kohlmeyer) PR #4760
• Bugfix for buffer overflows when using too many bonds, angles, dihedrals, or impropers per atom (Axel Kohlmeyer, Temple U) PR #4749
• Bug fixes and improvements for the GPU package (Trung Nguyen, U Chicago) PR #4764
• Update and improve support of bounding boxes for dynamic regions and fully support visualization of them in dump image/movie (Axel Kohlmeyer, Temple U) PR #4769
• Correct issues with generated CMake files and copied headers during "cmake --install" (Axel Kohlmeyer, Temple U) PR #4776
• Add instructions for GitHub Copilot code generation and code reivew (GitHub Copilot and Axel Kohlmeyer, Temple U) PR #4739
• Many small bug fixes, minor code improvements and refactoring, and updates for coding style, build system, documentation, and formatting (multiple authors): PR #4707, PR #4715, PR #4719, PR #4729, PR #4731, PR #4732, PR #4734, PR #4738, PR #4740, PR #4742, PR #4747, PR #4761, PR #4762, PR #4766, PR #4768, PR #4771, PR #4773, PR #4777, PR #4791, PR #4800, PR #4805

Backward compatibility notes:

• The fix addtoque command was renamed top fix addtorque/group to avoid confusion with the newly added fix addtorque/atom

---

Full Changelog: patch_10Sep2025...patch_10Dec2025

---

Information on the files attached to this release:

• lammps-src-10Dec2025.tar.gz - LAMMPS source code with translated manual in HTML and PDF format included
• lammps-linux-x86_64-10Dec2025.tar.gz - Fully static LAMMPS binaries for all Linux machines with x86_64 CPUs
• Manual_10Dec2025.pdf - LAMMPS manual as PDF
• lammps-gui-manual-v1.9.0.pdf - Manual for LAMMPS-GUI version v1.9.0 as PDF
• The following packages contain LAMMPS 10Dec2025 and LAMMPS-GUI v1.9.0 pre-compiled in different packaging formats
  • LAMMPS-Linux-x86_64-GUI-10Dec2025.tar.gz - Binaries with required libraries included using wrapper scripts. Compatible with Ubuntu 22.04LTS or later and equivalent Linux distributions
  • LAMMPS-Linux-x86_64-GUI-10Dec2025.flatpak - Linux binaries in flatpak bundle format and desktop integration
  • LAMMPS-macOS-multiarch-GUI-10Dec2025.dmg - DMG installer for macOS version 11 (Big Sur) or later. Includes support for Intel and Apple CPUs
  • LAMMPS-Win10-x64-GUI-10Dec2025.exe - Installer package for 64-bit x86 Windows 10 or later (may work on older Windows versions)

If you are looking for a pre-compiled LAMMPS-GUI version that can load a custom compiled LAMMPS shared library (version 22 Jul 2025 update 2 or later) please go to the LAMMPS-GUI release page

Update 2 for Stable release 22 July 2025

This is the second update to the 22 July 2025 stable release that fixes bugs that have been reported since its release and backported from the development version.

The following individual changes are included:

• make LAMMPS-GUI compatible with Qt-5.12 for building pre-compiled packages on Ubuntu 20.04LTS
• backport multiple LAMMPS-GUI bugfixes from upstream and step version to 1.7.2
• make CMake script code for generating headers with OpenCL kernels compatible with pathnames containing blanks
• fix off-by-one error when reporting multi-cutoff status
• fix bugs when reading angle, dihedrals, and impropers into a molecule template from JSON file
• improve confusing error message from read_dump during rerun
• error out when read_dump tries to create an atom with an out-of-range type
• fix bug in dump style extxyz that would prevent it from working with multiple MPI processes
• fix STORE/ATOM and STORE/GLOBAL ignore groups and thus are compatible with dynamic groups
• Fix incompatibilities with CUDA 13 runtime API in Kokkos lib (cherry-picked from upstream Kokkos, see kokkos/kokkos#8328)
• Fix bug in KOKKOS pair settings, where nargs was hardcoded to 1, which could mess up the cutoff distance if independent cutoffs were set for LJ and Coulomb
• fix incorrect use of address-of for 2d-arrays in AtomVecDielectric, AtomVecDipole, and AtomVecSpin. Only happens with general triclinic...
• reset eval_in_progress status when recreating variables
• fix data initialization indexing bug when overwriting internal style variables
• backport refactor from PR #4731 which includes a bugfix to add gravity during setup().
• backport of exchange communication buffer bugfix from PR #4749
• backport of bugfixes for RHEO, GRANULAR, and other packages from PR #4729 and PR #4766
• backport fixes for the GPU package from PR #4764
• don't provide bounding box information for variable shape or dynamic regions
• run some automated tests using GitHub runners also on "maintenance" branch
• Fix bug in fix langevin/kk
• Prevent race condition with Kokkos reverse comm and GPU-aware MPI

---

Full Changelog: stable_22Jul2025_update1...stable_22Jul2025_update2

---

Information on the files attached to this release:

• lammps-src-22Jul2025_update2.tar.gz - LAMMPS source code with translated manual in HTML and PDF format included
• lammps-linux-x86_64-22Jul2025_update2.tar.gz - Fully static LAMMPS binaries for all Linux machines with x86_64 CPUs
• The following packages contain LAMMPS 22Jul2025_update2 and LAMMPS-GUI v1.7.2 pre-compiled in different packaging formats
  • LAMMPS-Linux-x86_64-GUI-22Jul2025_update2.tar.gz - Binaries with required libraries included using wrapper scripts. Compatible with Ubuntu 20.04LTS or later and equivalent Linux distributions
  • LAMMPS-Linux-x86_64-GUI-22Jul2025_update2.flatpak - Linux binaries in flatpak bundle format and desktop integration
  • LAMMPS-macOS-multiarch-GUI-22Jul2025_update2.dmg - DMG installer for macOS version 11 (Big Sur) or later. Includes support for Intel and Apple CPUs
  • LAMMPS-Win10-64bit-GUI-22Jul2025_update2.exe - Installer package for 64-bit x86 Windows 10 or later (may work on older Windows versions)

If you are looking for a pre-compiled LAMMPS-GUI version that can load a custom compiled LAMMPS shared library please go to the LAMMPS-GUI release page. LAMMPS-GUI versions 1.8.0 to 1.8.3 are not compatible with this release, but versions 1.8.4 and later are.

Feature release 10 September 2025

This release starts the new development cycle of LAMMPS and implements some fundamental changes relative to the 22 July 2025 stable release:

• The minimum required C++ standard is raised from C++11 to C++17
• The minimum required CMake version is raised from 3.16 to 3.20
• GCC compilers older than version 9.x are known to have problems and are warned about
• The ATC, AWPMD, and POEMS packages are removed
• The traditional build system is restricted to packages that do not require extra build steps in the lib folder. The long-term goal is to phase this build system out and focus on building LAMMPS with CMake

Individual changes:

• Implement the changes mentioned above (Axel Kohlmeyer, Temple U) PR #4663
• Support single and mixed precision floating point in the KOKKOS package (Stan Moore, SNL) PR #4608
• Update the embedded Colvars library to version 2025-08-27 (Giacomo Fiorin, NIH and the Colvars developers) PR #4703
• Add "condense" option to delete_atoms, which compresses atom IDs similar to "compress", but preserves the order of atoms (Axel Kohlmeyer, Temple U) PR #4492
• Add support for rendering dynamic bonds to dump image (Axel Kohlmeyer, Temple U) PR #4686
• Add support for rendering regions to dump image (Axel Kohlmeyer, Temple U) PR #4696
• Support new dump image features in LAMMPS-GUI (Axel Kohlmeyer, Temple U) PR #4686 and PR #4696
• Unbundle LAMMPS-GUI from LAMMPS to its own repo and import as external project (Axel Kohlmeyer, Temple U) PR #4702
• Remove obsolete PyLammps and IPyLammps modules (Richard Berger, LANL) PR #4668
• Remove GJF implementation from fix langevin (Axel Kohlmeyer, Temple U) PR #4683
• Many small bug fixes, minor code improvements and refactoring, and updates for coding style, build system, documentation, and formatting (multiple authors): PR #4667, PR #4669, PR #4670, PR #4674, PR #4677, PR #4684, PR #4685, PR #4701, PR #4708, PR #4709, PR #4711

Backward compatibility notes:

• The requirements for C++17 and CMake 3.20 set the minimum Linux version to something like Ubuntu 22.04LTS and RHEL 8.x (with compiler upgrade, otherwise RHEL 9.x)
• Users of the ATC, AWPMD, or POEMS packages will have to use the 22 July 2025 version
• Users of the traditional build system will have to change to using CMake if they want to use packages like COLVARS, GPU, PLUMED, and several more that require build steps in the lib folder
• The PyLammps and IPyLammps modules are no longer available. Most of their functionality has been moved to the lammps module. See the manual for details.

---

Full Changelog: patch_22Jul2025...patch_10Sep2025

---

Information on the files attached to this release:

• lammps-src-10Sep2025.tar.gz - LAMMPS source code with translated manual in HTML and PDF format included
• lammps-linux-x86_64-10Sep2025.tar.gz - Fully static LAMMPS binaries for all Linux machines with x86_64 CPUs
• Manual_10Sep2025.pdf - LAMMPS manual in PDF version
• lammps-gui-manual-v1.8.3.pdf - LAMMPS-GUI manual for version v1.8.3 in PDF version
• The following packages contain LAMMPS 10Sep2025 and LAMMPS-GUI v1.8.3 pre-compiled in different packaging formats
  • LAMMPS-Linux-x86_64-GUI-10Sep2025.tar.gz - Binaries with required libraries included using wrapper scripts. Compatible with Ubuntu 22.04LTS or later and equivalent Linux distributions
  • LAMMPS-Linux-x86_64-GUI-10Sep2025.flatpak - Linux binaries in flatpak bundle format and desktop integration
  • LAMMPS-macOS-multiarch-GUI-10Sep2025.dmg - DMG installer for macOS version 11 (Big Sur) or later. Includes support for Intel and Apple CPUs
  • LAMMPS-Win10-64bit-GUI-10Sep2025.exe - Installer package for 64-bit x86 Windows 10 or later (may work on older Windows versions)

If you are looking for a pre-compiled LAMMPS-GUI version that can load a custom compiled LAMMPS shared library please go to the LAMMPS-GUI release page

Update 1 for Stable release 22 July 2025

This is the first update to the 22 July 2025 stable release that fixes bugs that have been reported since its release and backported from the development version.

The following individual changes are included:

• Lift restriction from fix neighbor/swap that limited the r0 parameter to integers
• Bug fixes for fix bond/react
• Bug fixes for BPM/RHEO/GRANULAR from PR #4685
• Fix bug in compute rdf which was ignoring user provided cutoff
• Fix bug when deleting computes that have created other computes
• Update external URLs and remove expired ones in the LAMMPS manual. Correct internal references.
• stop with an error in fix shake, if the topology data is inconsistent due to deleting only part of a molecule with shake constraints
• correct fix python/move example to be consistent with the current python module
• Allow access to select static properties in lammps_extract_global() without a LAMMPS instance
• change behavior of lammps.available_ids() for groups to always return the list of possible groups, since the list may have holes. This way the index in the list corresponds to the group index and can be used for tests against the per-atom "mask". This prevents segfaults and unexpected behavior
• refactor how fix shake operates during minimization. We cannot rely on the "closest" lists, as they seem to get out-of-sync. Instead determine bonds and apply forces directly. Let bond force return computed distance and thus collect restraint statistic directly
• bugfixes for the Colvars library from the Colvars developers (PR #4705)
• bugfixes for LAMMPS-GUI for bugs found during the LAMMPS workshop tutorial
• backport changes from LAMMPS-GUI 1.8.x to the 1.7.1 version included here
• update downloaded Plumed version to 2.9.4, support version 2.10.x if compiled externally.

---

Full Changelog: stable_22Jul2025...stable_22Jul2025_update1

---

Update 4 for Stable release 29 August 2024

This is the fourth update to the 29 August 2024 stable release that fixes more bugs that have been reported since then and updates some libraries, dependencies, tools, and build system support.

The following individual changes are included:

• fix issue in fix bond/react to require per-atom masses in molecule file when the system uses per-atom masses
• error out in compute msd if the number of atoms in the compute group changes by other means than a dynamic group
• avoid crash in utils::split_words() when processing partially quoted text
• fix bug in fix halt that would reset the timer timeout, even if it didn't trigger it
• fix bug in Fortran module where reallocation of global fix callback list did not happen as expected
• make pair styles (like EAM, MEAM, BOP, etc.) that call atom->set_mass() compatible with per-atom masses
• fix issue of yaml-cpp library included in ML-PACE failing to compile with GCC 15+
• bugfixes and LAMMPS-GUI optimizations and refactoring imported from upstream. It is now at version 1.7.0
• fix out-of-bounds access in initializing the cvatom array in pair.cpp, angle.cpp, dihedral.cpp, and improper.cpp
• updates of molecule command backported from upstream:
  • honor atom-id when parsing sections consistently
  • make parsing of header keywords more specific and thus don't misread incorrect ones
  • error out when the required "atoms" header keyword is missing
  • Masses section is required for body particles
• update the count of bodies when using create_atoms, fix deposit, or fix pour with a molecule template for body particles
• make python unit tests compatible with mpi4py version 4
• various fixes to classes where virtual and override were missing or used incorrectly as flagged by static code analysis
• various incorrect types or use of utils::numeric() instead of utils::inumeric() as flagged by static code analysis
• recognize a couple more Windows build numbers
• add caching for tpcpu and spcpu thermo keywords, so they can be used multiple times

Stable release 22 July 2025

Below is a list of major changes since the last stable release 29 August 2024

This stable release adds about 4500 commits since the previous stable release. This includes many contributions from beyond the core LAMMPS developers, but also continues the internal refactoring of the code base to make it more consistent and reliable yet also take more advantage of C++ features where they help to simplify the code and easier to modify, maintain, and contribute to.

This will be the last LAMMPS release that is compatible with C++11. Future releases will require C++17. Also, this will be the last release containing the ATC, AWPMD, and POEMS packages. These packages have been unmaintained for a long time and would require considerable effort to update to modern C++.

General Changes

• Continued refactoring of the core LAMMPS code and many packages to increase code reuse, simplify new additions, improve consistency, and benefit from C++11 features. Several convenience functions and classes were added in the process.
• Continued improving of error messages. For several common cases there is an indicator for which word caused the error. Also, if the command line had substitutions (e.g. from variables), the message will show both, the line as it appears in the input file and how it looks like after substitutions.
• Enhancements and extensions of the Programmer' Guide section of the manual.
• Significant enhancements to LAMMPS-GUI, performance improvements, and integration with https://lammpstutorials.github.io/
• Expanded CTest based internal testing.

Updates and new commands or styles or packages (for more details see the release note of the individual feature releases):

• Add support for running FFTs with NVPL from Nvidia for Grace CPUs for plain FFTs and KOKKOS
• Update support for using MKL FFTs for CPUs and add support for Intel GPUs when compiling LAMMPS with KOKKOS for SYCL
• Bugfixes, updates and ports for GRANULAR and related styles in KOKKOS
• Ports of several fixes to KOKKOS
• New coulomb pair style and QEq fix implementing CTIP as an alternative to could/streits
• New fix qtpie/reaxff that implements the QTPIE charge equilibration algorithm for ReaxFF
• Updates to the BPM and REHO packages
• Update embedded LAPACK compatible linear algebra functions to state of LAPACK 3.12.1
• New pair style dispersion/d3 to add D3 dispersion correction to DFT based ML potentials
• New angle style mwlc for (meltable wormlike chains)
• New fix style efield/lepton for more complex electric field calculations with custom expressions
• Discourage use of PyLammps and move its useful features to the regular LAMMPS python module (e.g. for use with Jupyter)
• Updates to fix imd to support the IMDv3 protocol and with it connect with MDAnalysis for on-the-fly processing
• Add MDR contact model to granular pair style
• Add capability to use atom style variables to pair style hybrid/scaled
• Add Kokkos versions of bispectrum (SNAP) and Gaussian descriptors on a grid
• new fix styles for bosonic path integral molecular dynamics
• new region2vmd command for visualizing LAMMPS regions with VMD
• new compute vacf/chunk command
• new extxyz dump style for writing extended xyz xyz files
• alternate velocity verlet timestepping option to avoid unphysical sticking in DEM simulations
• Improved molmap option for fix bond/react
• Improved error handling for bad/invalid arguments in the library interface, add function to stop Error class from printing errors directly, so they won't be printed twice when re-throwing them in the calling code
• Add support to move dipole moments along in rotate and transrot modes of fix move
• Add support for using fix adapt with dihedral and improper styles
• Updates for Moltemplate Howto
• Add fix gjf which was unbundled from fix langevin and provides significantly improved GJF Langevin dynamics
• Add pair style lj/pirani which is a generalized version of the LJ potential
• Add fix style ave/moments which can compute distribution moments in addition to averages
• Add fix set command
• Add support for writing more flexible equal-style and atom-style variables with Python functions
• Add support for reading JSON format molecule files
• Add support for writing plugins for run_style and minimize_style; fix multiple known bugs and refactor the whole framework to be a global and not a per-LAMMPS instance property
• Add atomic energy option for energy change calculation in fix sgcmc
• Updated bundled colvars library to version 2025-04-30
• Add fix neighbor/swap for hybrid MD/kMC-like simulations
• Add APIP package for "Adaptive Precision Interatomic Potentials"
• Add fix hmc for hybrid/Hamiltonian Monte Carlo
• Add molecule off/inter/intra option to fix bond/create to enable selecting only inter- or inter-molecular reactions

Backward compatibility notes:

• The bugfix change to the MS-MEAM in the meam pair style in PR #4331 changes energies and forces for systems using L12 lattice settings (like the example included in LAMMPS, or the force style unit test). The reference data for these has been updated.
• The new function utils::print() should be used instead of fmt::print()
• Backward compatibility with compute_modify extra and compute_modify dynamic was removed. Now using compute_modify extra/dof and compute_modify dynamic/dof is required.
• Using Python 2.x with LAMMPS should now cause an error, also when using Python 3.5 or older
• Compiling LAMMPS -DLAMMPS_SMALLSMALL will stop with an error, same for configuring with CMake
• When building the LAMMPS manual with "make fasthtml" we no longer use "Sphinx", but "pandoc". This produces an approximate output without many special Sphinx features, but can render most text very quickly and will only update files that were changed, while "make html" using Sphinx will always read and process all files.
• setting CMAKE_TUNE_FLAGS will have no effect anymore. Use CMAKE_CXX_FLAGS or CMAKE_C_FLAGS instead.
• using the gjf flag with fix langevin is discouraged. The new fix gjf should be used instead.

Stable release candidate 22 July 2025

This is the release candidate for the 2025 stable release of LAMMPS. It contains mostly bugfixes and cleanups plus a few features that didn't make it for the 12 June 2025 release.

Changes since the 12 June 2025 release:

• Update bundled Kokkos library to version 4.6.2 (Stan Moore, SNL and the Kokkos developers) PR #4580 and PR #4659
• Add KOKKOS version of fix electron/stopping (Ville-Markus Yli-Suutala, University of Helsinki) PR #4591
• Add fix neighbor/swap for hybrid MD/kMC-like simulations (Jacob Tavenner, NASA Ames Research Center) PR #4424 and PR #4615
• Add APIP package for "Adaptive Precision Interatomic Potentials" (D. Immel, R. Drautz, G. Sutmann, Jülich Supercomputing Centre and Ruhr-Uni Bochum), PR #4543
• Add fix hmc for hybrid/Hamiltonian Monte Carlo (Jack S. Draney and collaborators at Princeton U) PR #4019
• Add molecule off/inter/intra option to fix bond/create to enable selecting only inter- or inter-molecular reactions (William A. Pisani, US Army ERDC Environmental Laboratory) PR #4636
• Performance optimizations and bugfixes for KOKKOS (Trung Nguyen UChicago, Evan Weinberg Nvidia, Stan Moore SNL) PR #4573, PR #4601, #4638
• Bug fixes and performance improvements for LAMMPS-GUI which is now at version 1.7.0 (Axel Kohlmeyer, Temple U) PR #4646 and PR #4651
• Expanded unit tests (Axel Kohlmeyer, Temple U) PR #4645
• Many small bug fixes, minor code improvements and refactoring, and updates for coding style, build system, documentation, and formatting (multiple authors): PR #4627, PR #4628, PR #4630, PR #4633, PR #4634, PR #4637, PR #4642, PR #4649, PR #4650, PR #4653, PR #4654, PR #4655, PR #4656, PR #4657, PR #4658, PR #4660, PR #4661

Update 3 for Stable release 29 August 2024

This is the third update to the 29 August 2024 stable release that fixes more bugs that have been reported since then and updates some libraries, dependencies, tools, and build system support.

The following individual changes are included:

• Update LAMMPS-GUI to version 1.6.14
• fix issues scheduling computes on first step of dumps or when scheduled by time or variables
• make finding libcurl work with older Linux versions and support fully static builds on Linux using MUSL cross-compiler
• fix OpenMP bug in pppm/dielectric and pppm/disp/dielectric
• don't mix malloc()/free() and new/delete in pair styles mliap and mliap/kk
• plug several memory leaks when using pair style mliap/kk
• fix issues with pair style mliap unified and mliap/kk unified
• plug memory leaks when calling ev_init() directly or indirectly from KOKKOS styles
• fix multiple issues in pair styles dpd/coul/slater/long and dpd/coul/slater/long/gpu
• fix indexing bug for optional arguments of compute reduce and compute reduce/region
• error out or ignore displacement in z-direction for 2d systems
• fix issue in reference data for KIM package extended tests
• bugfix for intensive/extensive settings for fix electron/stoppin/fit
• bugfix for velocity remap with fix deform
• bugfix for EAM pair styles in the GPU package when compiling with OpenCL
• bugfix for segfaults when loading a kspace style from a plugin
• work around consteval issue in bundled {fmt} library with compilers requesting C++20 or later
• accept denormal floating-point numbers when converting with ValueTokenizer::next_double() or utils::numeric()