Jonathon Misiewicz
Jonathon Misiewicz
A valuable idea, but per conference call, this is the kind of quality of life improvement we're deferring to 1.5.
Please request review from me once both #2663 is in and all merge conflicts are resolved.
My one request is that variables that are obviously used for indexing/lengths be `size_t`.
Are we clear to remove those four tests, then?
For reference, this issue is _also_ known to cause problems in geometry optimizations. You read in the orbitals from the previous geometry but don't correct for orthogonality. Fock matrix diagonalization...
I'd prefer a canonical or symmetric orthogonalization scheme to Gram-Schmidt, as the Gram-Schmidt vectors are dependent on how you order them. But yes, the consensus in this topic was already...
> Thanks, the basis set name makes sense. I will start naming my basis sets, but would it be possible to use a deterministic name for basis sets instead of...
How does this PR differ from #2642?
The action item here is to have some way to adjust the symmetrization threshold.
You said it's a doublet in the molecule input. Psi should honor that. This is a bug.