Releases: metatensor/featomic
featomic-torch v0.7.3
This is a small patch release of featomic-torch, adding support for PyTorch v2.10 in the prebuilt wheels.
featomic-torch v0.7.2
This is a patch release for featomic-torch, with the following changes:
- We now require Python >= 3.10
- Added support for torch v2.9, removed support for torch 2.1 and 2.2
featomic v0.6.4
This is a patch release, adding support for metatensor-operations v0.4
featomic v0.6.3
This is a small follow-up release to 0.6.2, fixing the dependency constrain on metatensor.
featomic v0.6.2
This is a patch release of featomic updating metatensor to v0.1.17, which should bring faster Labels creation and fix compilation with recent rust compilers.
featomic-torch v0.7.1
This is a patch release of featomic-torch, adding support for torch v2.8 and metatensor-torch v0.8.0.
featomic-torch v0.7.0
This version switches metatensor.torch.atomistic to its replacement in metatomic.torch.
featomic-torch v0.6.2
Added support for PyTorch 2.7, removed support for PyTorch 2.0 and older.
featomic-v0.6.1
This is a patch release of featomic
What's Changed
Added
clebsch_gordan.EquivariantPowerSpectrumByPaircalculator for two-center equivariant
descriptors. Similar API toclebsch_gordan.EquivariantPowerSpectrum(#378)
Fixed
- Fixed
featomic.clebsch_gordan.cartesian_to_sphericaltransformation for tensors of
rank greater than 2 (#371) - Fixed the calculation of Clebsch-Gordan coefficients, used all across the
featomic.clebsch_gordanmodule (#371)
Full Changelog: featomic-torch-v0.6.1...featomic-v0.6.1
featomic-torch-v0.6.1
This patch release adds support for Python 3.13 and PyTorch 2.6 (#384)