ROHF conventions

Reading https://github.com/cclib/cclib/pull/298, I realized there aren’t any conventions that I know of for what cclib does in the case of ROHF calculations. Is this true?

These are results for DVB cation, ROHF rather than UHF:

Attempting to read dalton.out
homos:
[34]
Attempting to read gamess.out
homos:
[34 33]
Attempting to read gaussian.out
homos:
[34]
Attempting to read molpro.out
homos:
[34 33]
Attempting to read nwchem.out
homos:
[34]
Attempting to read orca.out
homos:
[34]
Attempting to read qchem.out
homos:
[34 33]

No Psi4 result because it doesn’t handle UHF/ROHF right now, or the latest version of the program; I’m making major changes to the parser and test files.

I think we should have unit tests for restricted open-shell calculations, which should be defined as (to start with):

assert len(data.homos) == 2
assert len(data.mocoeffs) == 1
assert len(data.moenergies) == 1

[langner]

I think Eric’s previous proposal to use a list of integers is the best one.

[berquist]

I’m thinking two stages:

• one PR to implement the fixes and add either unit or regression tests, not sure which is best, but leaves homos as-is
• another PR that handles the homos <-> nelectrons inversion