MBIS output written twice with differing atom numbers in 1.4rc(2,3)

1

Hi there,

the mbis-1 test with psi4 1.4rc2/3 actually has now two

==> Computing MBIS Charges <==

sections with different numbers of atoms in the text output. The test, however, passes.

This is somewhat confusing for those who parse the numbers directly from the text output file.

1.4rc1 does not show this behavior.

Is it a bug, or actually intended behavior?

Regards,

Udo

2

@jeffschriber, I’m pretty sure this was your code.

3

This is intended behavior, though admittedly fairly confusing. One of the properties returned by the MBIS routine is the volume ratios, which takes each MBIS atomic volume and divides it by a corresponding free-atom volume (i.e., it gives info on how much the atomic volume changes when you take a free atom and stick it in a molecule). In Psi, we compute these free-atom volumes on-the-fly in order to match whatever level of theory is used for the molecule (with UHF/UKS as needed). So, there is a lot of extra printing in the output for the free-atom computations, which should account for those sections with different numbers of atoms (should be just single atoms). The MBIS summary printed at the end of the output that lists all of the atoms is correct, and should have an additional “Volume Ratios” field that not present for the printed single-atom computation.