Minus Multiplicity for Fragments in opposite spin coupling and in SAPT0(UHF)


I’m trying using PSI4 to constructe a complex state where the local spin of different parts in the system are antiferromagnetcally coupled.

According to the experience from using other softwares, the multiplicity for different parts could be set with opposite sign to represent the antiferromagnetic coupling. And, for the system I’m handling with, guess=mix wouldn’t be sufficient to get the wanted wavefunction due to the complexity of the coupling.

I accordingly tried fragment description for the molecule in PSI4, only to find there will be errors when set the fragment multiplicity to be minus values like -2.
Fatal Error: Illegal atom symbol in geometry specification: 0 on line
0 -2
while the error will disappear if no minus multiplicity presents.

Is minus multiplicity supported currently, or is it possible to make a equivalent set in PSI4?
Besides, will SAPT0(UHF) support the broken-symmetry like dimer, such as one will encounter in the dissociation of H2, if Fragment had supported the minus multiplicity?

Thanks for your patience and looking forward to your reply.

Bset regards,


Probably a question for @jgonthier.

As far as I know, Psi4 does not support negative multiplicities for fragments. That means that, to my knowledge, Psi4 always takes positive multiplicites and the excess unpaired electrons always have alpha spin.

If Fragments were supporting negative multiplicites, and attributing beta spin to corresponding electrons, then I believe SAPT0(UHF) would support the broken-symmetry dimer. It would not, however, introduce any spin coupling between the dimers. As a side note, if the dimers are strongly interacting, be aware that low-order perturbation theory (such as SAPT0) might fail to give a meaningful answer.

If negative multiplicities get implemented into Psi4, I’ll be happy to make sure the SAPT0(UHF) code can support them.

Thank you for your kind help and hope the feature can be supported in the future.
I‘ll be careful when handling with cases described in the reply.
All the best with PSI4.