open-shell. Although optking doesn't do the electronic structure
and so doesn't use multiplicity etc., it sends such information
through qcengine to external programs. There is some ambiguity
as to whether a user would typically want us to use those fragments
in our representation and optimization. In special cases of gradient
calculations involving multiple, unique fragments, it might
be easier to use the OptHelper class and feed gradients in that way.
[x] First pass successfully does anions and triplets.
[ ] Need to test serialization during/between steps to make sure we can export/import our molsys.
[ ] Need to think through how to handle or warn about cases where the external gradients depend on things like fragment multiplicity but internally we are doing single molecule (for example).
open-shell. Although optking doesn't do the electronic structure and so doesn't use multiplicity etc., it sends such information through qcengine to external programs. There is some ambiguity as to whether a user would typically want us to use those fragments in our representation and optimization. In special cases of gradient calculations involving multiple, unique fragments, it might be easier to use the OptHelper class and feed gradients in that way.