v0.4.0

What's Changed

💥 Breaking Changes

• Change default CHGnet.load(check_cuda_mem: bool) to False by @janosh in #164

🐛 Bug Fixes

• Fix KeyError: 'decay_fraction' and TypeError: Object of type int64 is not JSON serializable by @janosh in #169
• Fix missing create_graph.c in source distribution by @DanielYang59 in #201
• Remove reverse readline, test again NP1 and recover NumPy 1 dependency support by @DanielYang59 in #203

🛠 Enhancements

• Add wandb logging support to Trainer class by @janosh in #166
• Add keyword wandb_log_freq: LogFreq = LogEachBatch to Trainer.train() by @janosh in #170

📖 Documentation

• Dispersion by @ajhoffman1229 in #192

🧹 Linting

• ruff fixes by @DanielYang59 in #184

🏷️ Static Typing

• Use return type typing_extensions.Self for class methods by @janosh in #179
• Create py.typed by @Andrew-S-Rosen in #189

🏥 Package Health

• Support NumPy 2 by @DanielYang59 in #202
• Drop Python 3.9 support by @janosh in #204

New Contributors

• @ajhoffman1229 made their first contribution in #192
• @DanielYang59 made their first contribution in #201

Full Changelog: v0.3.8...v0.4.0

v0.3.8

Important

v0.3.8 is a hot-fix release to replace v0.3.7 which we plan to yank from PyPI due to lack of compiled wheels, resulting in installation issues as reported in #160

What's Changed

🐛 Bug Fixes

• Fix release CI by @janosh in #161

Full Changelog: v0.3.7...v0.3.8

v0.3.7

What's Changed

📖 Documentation

• Fix eslint by @janosh in #151

💥 Breaking Changes

• ruff fixes + type annotations by @janosh in #156

🏥 Package Health

• Fix "Failed to CreateArtifact" on release by @janosh in #152
• Bump min ASE to 3.23.0 by @janosh in #159

Full Changelog: v0.3.6...v0.3.7

v0.3.6

What's Changed

🛠 Enhancements

• Fixing VASP parsing with no MAGMOM by @BowenD-UCB in #148

🚀 Performance

• Allowing avoiding sorting cuda memory, enhancing compatibility working on HPC by @BowenD-UCB in dd0dd07

Full Changelog: v0.3.5...v0.3.6

v0.3.5

What's Changed

🛠 Enhancements

• Add assign_magmoms=True keyword to StructOptimizer.relax() by @janosh in #124

🚀 Performance

• Use MPS backend if available and use_device=None, add CHGNET_DEVICE env var by @janosh in #131

🏥 Package Health

• Add Python 3.12 to officially supported versions and test in CI by @janosh in #139

🤷‍♂️ Other Changes

• Use new uv package manager to install deps in CI by @janosh in #133

Full Changelog: v0.3.4...v0.3.5

v0.3.4

What's Changed

🛠 Enhancements

• Better backward compatibility with ase versions
• Better backward compatibility with pymatgen versions
• Allowing loss_ratio=0 for Trainer

🐛 Bug Fixes

• Add pip install git+https://gitlab.com/ase/ase user advice on FrechetCellFilter ImportError and allow ase_filter to be str by @janosh in #104
• Fix solve_charge_by_mag() using wrong key in structure.site_properties.get('final_magmom') by @janosh in #114

🧹 House-Keeping

• Drop __all__ by @janosh in #106

🏥 Package Health

• Fix release CI by @janosh in #122

🤷‍♂️ Other Changes

• Add keyword use_device to CHGNet.load() by @tsihyoung in #105
• Update dynamics.py by @zhongpc in #109
• cuda_devices_sorted_by_free_mem() return [] if not torch.cuda.is_available() by @janosh in #115

New Contributors

• @zhongpc made their first contribution in #109

Full Changelog: v0.3.3...v0.3.4

v0.3.3

What's Changed

🐛 Bug Fixes

• Replace ase ExpCellFilter with FrechetCellFilter in StructOptimizer by @janosh in #101

If you encounter ase.filters import issue, please install the newest ase from their source before the next ase release:
pip install git+https://gitlab.com/ase/ase

🛠 Enhancements

• Add ase_filter keyword to StructOptimizer.relax() by @janosh in #102

Full Changelog: v0.3.2...v0.3.3

v0.3.2

Changes

• Link to Video tutorial @BowenD-UCB
• Allow setting MD start temperature @BowenD-UCB
• Fixed bug triggerred by dtype @janosh in #95

Full Changelog: V0.3.1...v0.3.2

v0.3.1

Hot fix release for v0.3.0 which is deprecated due to missing pretrained checkpoints (fixed in 0f02786). See #86 for details.

🛠 Enhancements

• Add CHGNet.version property by @janosh in #86

Full Changelog: V0.3.0...v0.3.1

v0.3.0

0.3.0 version: Improved pretrained weights released

We release the most recent pretrained model: CHGNet 0.3.0 🎉 (see details)
CHGNet.load() now defaults to '0.3.0' version.

Major changes:

1. Increased AtomGraph cutoff to 6A #55
2. Resolved discontinuity issue when no BondGraph presents #79
3. Added some normalization layers
4. Slight improvements on energy, force, stress accuracies

Full Changelog: v0.2.2...v0.3.0