Build a wfn object from a qcengine results wavefunction dict

[hadim]

I am trying to generate .cube files using the output of qcengine.compute():

results = qcengine.compute(...)

wfn_dict = results.wavefunction.dict()

The issue is that wfn_dict is incompatible with the psi4 wfn dict I could reload with psi4.core.Wavefunction.from_file(wfn_dict).

Then I could simply do:

psi4.cubeprop(wfn)

See also MolSSI/QCElemental#248

[JonathonMisiewicz]

What exactly is the issue here? I'm not familiar with QCEngine, but as I understand it, your complaint is that qcengine.AtomicResult.wavefunction doesn't contain all the information that Psi4 needs for cubeprop. This sounds like something for the QCEngine developers to work on, not the Psi4 developers.

[hadim]

I am actually more looking for advice than complaining or reporting a bug here. It's indeed more of a qcengine thing, but I was wondering whether psi4 had a way to perform the conversion.

I spotted psi4.schema_wrapper.driver._convert_wavefunction() (

psi4/psi4/driver/schema_wrapper.py

Line 493 in 6260856

"Using `psi4.schema_wrapper.run_json` or `psi4.json_wrapper.run_json` instead of `psi4.schema_wrapper.run_qcschema` is deprecated, and as soon as 1.5 it will stop working\n",

) that can convert a psi4_wfn to a qcschema_wfn and I was wondering whether a function to perform the inverse operation exists somewhere (doing the qcschema_wfn to psi4_wfn conversion).

I totally understand this is not strictly a psi4 issue here so I'll close if you think it's irrelevant.

[loriab]

"Round-tripping" of wfn (psi4.core.Wavefunction -> AtomicResult.wavefunction -> psi4.core.Wavefunction) is a definite goal, but you concluded rightly that it does not work now. A needed step is simple round-tripping of BasisSet, and my last bout of working on that got to this stage: #2861 . I'd like to work on this again, but I don't know when.

It's fine to leave this open, in my opinion, since this is a question others may have.

[loriab]

Sorry, that #2861 link was an error. MolSSI/QCElemental#230 was what I meant.