I’m trying to do EOM-CC3 calculations for the benzene molecule (input and output are attached), but the CC3 code calculates the correction (from EOM-CCSD) for just one state. I want to calculate the CC3 correction for all roots. Can I fix this? Or does the code calculate one correction at a time?

In addition, does the psi4 code have another approximation instead of CC3 for triples correction in excited states?

The nature of CC3 forces the algorithm to compute one root at a time because we must avoid storing triple excitation vectors. CC3 is the only excited-state triples method in Psi4.

The EOM code first finds the number of EOM-CCSD roots requested and then, by default, uses the last state in the current irrep to follow for the CC3 correction.

If you wish to compute the CC3 energy for a state that is not the highest energy state computed with EOM CCSD, you should be able to select it with the “PROP_ROOT” keyword. Occasionally this is helpful for convergence of the EOM CCSD or convenient to be able to see what states are just higher.