What's Changed

• Remove for loop over multiplicities in D_from_angles by @SimonWagnerHD in #517
• Add metadata for linear slices by @mitkotak in #518

New Contributors

• @SimonWagnerHD made their first contribution in #517

Full Changelog: 0.5.7...0.5.8

What's Changed

• Fix soft_unit_step for torch.compile. by @ameya98 in #506
• Compile on by default by @mitkotak in #497
• Added e3nn.math.bessel by @mitkotak in #508
• Porting o3.Irreps.filter from e3nn-jax by @mitkotak in #515
• Porting slice_by_mul from e3nn-jax + bug fix in filter by @mitkotak in #516
• version bump to 0.5.7 by @mitkotak in #514

New Contributors

• @ameya98 made their first contribution in #506

Full Changelog: 0.5.6...0.5.7

What's Changed

• Refactor by @cw-tan in #502
• 0.5.6 version bump by @mitkotak in #504

New Contributors

• @cw-tan made their first contribution in #502

Full Changelog: 0.5.5...0.5.6

What's Changed

• bumping version to 0.5.5 by @mitkotak in #500

Full Changelog: 0.5.4.1...0.5.5

What's Changed

• hack to turn off jit script in _spherical_harmonics by @mitkotak in #485
• Fixes #493 by @mitkotak in #494
• explicitly specify device placement in random number generator in e3nn.math._normalize_activation.moment by @lyuwen in #492
• Force CodeGenMixin to entirely respect e3nn.set_optimization_defaults by @Linux-cpp-lisp in #484
• Replace flake8 with ruff by @mitkotak in #498

New Contributors

• @lyuwen made their first contribution in #492

Full Changelog: 0.5.4...0.5.4.1

What's Changed

• PT2 compile with Legacy code by @mitkotak in #455
• Release fix by @mitkotak in #469
• update deprecated matplotlib option by @eszter137 in #475
• Fix docs build by @mitkotak in #476
• Updating PyPI token in release process by @mitkotak in #478
• updated citations format to make twine happy by @mitkotak in #479
• Bump python version to only 3.11 in tests by @mitkotak in #487
• Small change to pyproject.toml to fix pypi docs by @rmcconke in #488
• PyPI release 0.5.4 by @mitkotak in #489

New Contributors

• @eszter137 made their first contribution in #475
• @rmcconke made their first contribution in #488

Full Changelog: 0.5.2...0.5.4

[0.5.2] - 2024-07-26

Added

• o3.experimental.FullTensorProductv2 | ElementwiseTensorProductv2 for compatibility with torch.compile(..., fulgraph=True)
• enable pip caching in CI
• Optional scalar bias term in _batchnorm.py

Changed

• refactor to use pyproject.toml for packaging
• refactor gh community files
• move pylint, coverage and flake8 configuration to pyproject.toml

Fixed

• Fix TorchScript warning "doesn't support instance-level annotations" (#437)

Added

• L=12 spherical harmonics

Fixed

• TensorProduct.visualize now works even if the TP is on the GPU.
• Github actions only trigger a push to coveralls if the corresponding token is set in github secrets.
• Batchnorm

[0.5.0] - 2022-04-13

Added

• Sparse Voxel Convolution
• Clebsch-Gordan coefficients are computed via a change of basis from the complex to real basis. (see #341)
• o3, nn and io are accessible through e3nn. For instance e3nn.o3.rand_axis_angle.

Changed

• Since now the code is no more tested against torch==1.8.0, only tested against torch>=1.10.0

Fixed

• wigner_3j now always returns a contiguous copy regardless of dtype or device

[0.4.4] - 2021-12-15

Fixed

• Remove CartesianTensor._rtp. Instead recompute the ReducedTensorProduct everytime. The user can save the ReducedTensorProduct to avoid creating it each time.
• *equivariance_error no longer keeps around unneeded autograd graphs
• CartesianTensor builds ReducedTensorProduct with correct device/dtype when called without one

Added

• Created module for reflected imports allowing for nice syntax for creating irreps, e.g. from e3nn.o3.irreps import l3o # same as Irreps("o3")
• Add uvu<v mode for TensorProduct. Compute only the upper triangular part of the uv terms.
• (beta) TensorSquare. computes x \otimes x and decompose it.
• *equivariance_error now tell you which arguments had which error

Changed

• Give up the support of python 3.6, set python_requires='>=3.7' in setup
• Optimize a little bit ReducedTensorProduct: solve linear system only once per irrep instead of 2L+1 times.
• Do not scale line width by path_weight in TensorProduct.visualize
• *equivariance_error now transforms its inputs in float64 by default, regardless of the dtype used for the calculation itself