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

v0.2.2

Important Notes

v0.2.2 has a breaking change in model initialization from scratch i.e. chgnet = CHGNet(), to address #79.
The developer team is running another round of pretraining and the updated weights will be released in the next release.

For now, chgnet = CHGNet.load() loads same pretrained model as previous releases.
If user wants to load their own fine-tuned weights trained from previous releases, please do
chgnet = CHGNet.from_file("your_weights_path", mlp_out_bias=True)

What's Changed

• Run cibuildwheel on Apple silicon to publish pre-compiled macOS ARM binaries by @janosh in #78
• always report cov by @lbluque in #77
• 1.Dropped torch.det for MPS compatibility 2.better testing for rotation/supercell invariance by @BowenD-UCB in 4cee9e6
• Added function to output site energies by @BowenD-UCB in 77845dc
• Added examples for parsing DFT by @BowenD-UCB in 2ebc57f
• Fixed bug in #79, while not breaking loading pretrained weights by @BowenD-UCB in 1a558ea
• CHGNet now supports Apple MPS:tada:, pytorch-MPS has been tested with torch-2.0.1 by @BowenD-UCB in 516c422
• added num_isolated_atoms property by @BowenD-UCB in 20ebc58
• allow specifying decay fraction in CosineAnnealing LR scheduler by @BowenD-UCB in dbb0305
• Moved all torch.int64 to torch.int32 for MPS support by @BowenD-UCB in 828478a
• added notes to avoid Colab bug by @BowenD-UCB in a760ec8
• added installation from source by @BowenD-UCB in 030d897
• removed print message by @BowenD-UCB in 38a5cc4
• minor fix by @BowenD-UCB in 465f833
• fixed test by @BowenD-UCB in d5ba24b
• loosen numerical tolerance by @BowenD-UCB in b59d301
• added gMLP_norm option if gMLP is used at read_out by @BowenD-UCB in 195d48c

Full Changelog: v0.2.1...v0.2.2