Posted by Ignacio Fdez. Galvan on September 24, 2012 at 16:48:26:
In Reply to: CASPT2 energy calculation result is different from Molcas62 and Molcas74 posted by Dongwen Zhang on September 24, 2012 at 14:59:58:
A new zeroth-order Hamiltonian for CASPT2 was introduced in Molcas 6.4 if I remember correctly. This is supposed to reduce systematic errors introduced in open-shell configurations. (See Chem. Phys. Lett. (2004) 396, 142-149.)
You can get the old results with new Molcas versions by adding "IPEA 0.0" to your CASPT2 section.