Guillaume Fraux

This would mainly require copying the code from here to a Python function: https://github.com/Luthaf/rascaline/blob/5ac2ee9e6b45f31ba608c02d6d8dd7d82b2009a9/rascaline/src/math/splines.rs#L228-L284

Yes, rascaline currently only supports calculations on CPU. All the CPU-intensif code is written in Rust, and then exported with a compatibilty shim to PyTorch, but the main calculation happens...

I split the GPU support discussion in a separate issue (#271). ---- Regarding the output format: rascaline uses [metatensor](https://github.com/lab-cosmo/metatensor) to store it's output, the main reason for this is that...

> to get the Wigner-D matrix of rascaline as follows(the rotations matrix are randomly generated and do not correspond to the values in the image above). Could you share the...

- [ ] function that compute metadata should not need to allocate memory for the data arrays. We should have a new `EmptyArray` class in Python's metatensor API that just...

- [x] the current tests are quite slow, we should see how to make them faster! ```bash tox -e all-deps -- python/rascaline/tests/utils/clebsch_gordan.py 27 passed in 77.57s ``` This should not...

Could you please explain what you mean by "density reconstruction" here? Building the atomic density from the spherical expansion coefficients? And "doesn't work" here means you have a high error...

So the issue here is that the `GtoBasis` class computes the > Primitive (not normalized nor orthonormalized) GTO radial basis. > > https://luthaf.fr/rascaline/latest/references/api/python/utils/radial-basis.html#rascaline.utils.GtoBasis But the spherical expansion uses orthonormalized GTO...

Couple of points: we already have code to do this in Rust: https://github.com/Luthaf/rascaline/blob/55ad1f14363812639c865801a6d517d8c5f363f6/rascaline/src/calculator.rs#L288. We mainly need to expose this code to the C and then Python API. Regarding jax, it...

This is solved for Python, but we still need an example for the Rust, C and C++ API