bizarre CCSD correlation energy


Click here for full thread
Clicked A Few Times
Hello,
it looks to me (I may be wrong) that this may be a case of switching between various UHF solutions (or HF instability). Have you tried to "continue" the UHF solution from c1 to c2 and then perform CCSD(T) calculations on the resutling UHF reference?

In order to do so, please move from c1 to c2 geometry using small "geometry increments" and for
next geometry use the converged UHF vectors from the previous geometry. To do so please use the
"vectors input" "vectors output" directives described in the HF section of the NWChem manual. You don't have to perfrom CCSD(T) calculations for each intermediate geometry. The point is to perform one-step CCSD(T) calculations for c2 geometry using UHF reference obtained in such a way.
Best regards,
Karol