Dirac interface to MRCC - working for anyone?

We back to trying to get the Dirac to MRCC interface working, particularly for 4-component coupled cluster (higher than (T)) calculations. Currently even closed-shell CCSD is not coming close to reproducing the CCSD energy from Dirac. Using Dirac21, the SCF energy matches exactly (Dirac23 does not accomplish this), but the CCSD energy in the first iteration in MRCC increases by 4 Hartrees (and of course doesn't converge if allowed to run). We're using the 2023 version of MRCC. 

Last time, we tested the interface with Dirac12. With that version, it should work.

wow, that was a long time ago. We'll see if we can get that version up and running.

best regards, -Kirk

Quick update - we've found that the interface currently works for Dirac18, but not for versions any newer than that. I'll post an update if we can easily get it working for the latest version.

Some weirdness I hope you can help with. With Dirac18, closed shell cases work more or less. Unlike the Be example you show in the manual, it only works if one steps down to D2h symmetry. Keeping Dinfh symmetry gives the error:
Invalid spatial symmetry!

modifying the refdet vector in MINP for an open-shell atom (Be+ for example) always gives this invalid spatial symmetry error.

Ideas?

best regards, -Kirk

The program works only with Abelian point groups, so you should use D2h for atoms.