Skip to content

v0.1.0

Latest
Latest

Choose a tag to compare

View all tags
@wanghan-iapcm wanghan-iapcm released this 18 Jun 07:40
v0.1.0
07d8903
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

First release since v0.0.7 (Jan 2023). This is a large feature release spanning ~2.5 years of development, including new exploration engines, first-principles backends, training enhancements, and significant workflow/performance improvements.

⚠️ Breaking changes & requirements

  • Minimum Python is now 3.9 (raised from 3.7) (#336)
  • deepmd-kit v3 support — training/inference now targets dp v3 (#207, calypso compat #201)
  • Removed support for the old-style input (#176)
  • Removed support for Lebesgue integration (#142)
  • Removed set_prefix (#219)

🚀 New exploration engines

  • DiffCSP as a new exploration engine (#251)
  • CALYPSO structure-prediction support, with model-deviation refactor for speedup (#188, #217, #224, #197)
  • PIMD support for LAMMPS (#263, #266)

🔬 First-principles / labeling

  • CP2K support (#238)
  • ABACUS support via fpop (#174, #218)
  • Gaussian: extra FP outputs and post_command (#256)
  • Electronic temperature support in FP and LAMMPS NPT task groups (#258, #255)

🧠 Training

  • Finetune on pretrained models, plus finetune refactor and resubmission support (#152, #247, #165)
  • Knowledge distillation (#122)
  • Mixed-type training data (#145)
  • Multitask training: validation data support; split last-iteration data into train/valid (#257, #293)
  • Configurable train args and optional train files (#233, #236)

⚙️ Workflow, performance & I/O

  • HDF5 support for trajectories and model-deviation; RunLmpHDF5 (#259, #267)
  • Workflow parallelism (#222) and conf-filter parallelization (#268)
  • dflow debug mode for local runs (#125)
  • Rewritten download implementation with multiple fixes (#135, #148, #149)
  • URI support for init_data, multi_init_data, valid_data, init_model (#227)
  • Memory/space optimizations for ExplorationReport (#279, #262)

🧩 Configuration generation & filtering

  • Distance conf filter; DistanceConfFilter optimized and BoxSkewnessConfFilter fix (#250, #292)
  • input_extra_files for lmp-md and lmp-template (#306)
  • Customized LAMMPS/PLUMED template task groups (#156)

🖥️ CLI / GUI

  • dpgui entry point and dpgen2 gui command (#172)
  • dpgen2 status reads/writes the exploration scheduler via global outputs (#129)

📅 Scheduling & reports

  • Adaptive lower trust level (#118)
  • Candidate-selection probability style inv_pop_f (#157, #300)
  • Model-deviation management in the report class (#143)

📚 Docs & examples

  • Guide: "Implement a new exploration engine for dpgen2" (#248)
  • New CP2K, CHNO, and Slurm examples (#158, #291)
  • Docs theme switched to sphinx-book-theme (#270)

🛠️ Maintenance & CI

  • Adopted ruff formatter, isort, and pre-commit (#181, #126, #120)
  • Dependabot for GitHub Actions; numerous action bumps
  • Docker base fixed to python:3.11-slim + build-time git, with a build-only PR check (#345)
  • Test fix for the Masses section in dpdata ≥ 1.0.2 lammps output (#343)

Full changelog: v0.0.7...v0.1.0

Assets 2
Loading