Partial ECPs
Closed this issue · 3 comments
I am not sure how to handle the case in which I have ECPs only for certain atoms in my system. For example, I am testing with the water molecule, and I have an ECP for the 1s of oxygen, but a normal basis set for H. So I need to populate the ECP group, however, there are certain entries that may be ambiguous, for example max_ang_mom_plus_1
? Or in the end, because in nucleus_index
there will be no entry pointing to the hydrogen indices, everything works out?
PS: I am collecting a list of comments/possible feature additions while implementing this interface that I will share when I finish. One of the feature addition could be the support for fully non-local pseudopotentials, such as the ones used in CP2K (separable dual space norm-conserving GTH pseudopotentials).
Yes, nucleus_index
is your friend here, so you will have no entries for H. The entries for H in max_ang_mom_plus_1
will be ambiguous but we can probably agree to put some value for cases like yours. In the end, it should not matter if the code takes the proper use of nucleus_index
array.
By the code
I mean the user code, which reads TREXIO format and converts into code-specific representation
Alright, indeed that is what I have done. I initialize max_ang_mom_plus_1
to all zeros, and only the relevant atoms will have something greater than zero.