Luthaf / rascaline

Computing representations for atomistic machine learning
https://luthaf.fr/rascaline/
BSD 3-Clause "New" or "Revised" License
44 stars 13 forks source link

Make sure the central atom is considered it's own neighbor in spherical expansion #91

Closed Luthaf closed 2 years ago

Luthaf commented 2 years ago

Related to #90, there might be more to it.

kvhuguenin commented 2 years ago

I have now added the global factor of 4pi that was missing from the center contribution to the relevant section of the code. This solves the two issues I had mentioned in #90. @Luthaf If you also want to run new sanity checks on your own, the rascaline and librascal expansion coefficients should now agree up to a global factor of 26.697 (the librascal coeffs are smaller by this factor, so taking the inverse, they are .037456 times the pyLODE/rascaline coefficients).