Change Default Behavior of "fix_com" and "fix_orientation" about qcelemental HOT 6 CLOSED

It's a lot easier to detect and exploit symmetry when molecules are in the "standard orientation", which reduces cost. It also just generally makes things consistent (in the standard orientation, principal/highest order rotation axis gets aligned with the Z-axis).

See for example, Gaussian: https://gaussian.com/symmetry/

from qcelemental.

If using QCEngine and I set fix_com and fix_orientation to True will this prevent all the various harnessed programs from doing these things?

It's supposed to, but testing rolled out after several of the harnesses, so I can only vouch for the ones I worked on and tested: https://github.com/MolSSI/QCEngine/blob/master/qcengine/programs/tests/test_alignment.py#L88-L95

from qcelemental.

I agree that it can be surprising -- I've gotten burned myself. A lot of QC programs do the same by treating input cartesians as eligible for reorientation. I think it's a case of either default will surprise in a good fraction of cases, so best to leave as-is. It wouldn't be too hard to switch here, but a lot of QCEngine logic would change.

That said, the description you quoted about QCEl pre-processing the Mol doesn't ring quite true to me (though I could be mistaken). It sounds more like what QCPortal might do rather than the Mol constructor alone.

from qcelemental.

For my own edification, why do "a lot of QC programs do the same by treating input cartesians as eligible for reorientation"? I don't actually know why this would be considered desirable--what's the intention behind doing this?

from qcelemental.

Another thing to watch out for is that some QC programs shift to center of mass, and some shift to center of charge (so that isotope substitutions use the same frame).

from qcelemental.

Yikes! Thanks for the heads up!

If using QCEngine and I set fix_com and fix_orientation to True will this prevent all the various harnessed programs from doing these things?

from qcelemental.