Posted by Björn Roos on June 02, 2004 at 07:59:36:
In Reply to: Re: CASPT2 analytic gradient was so bad??? posted by Evgeny K. Dolgov on May 17, 2004 at 13:41:16:
: Great thanks for all who answered me!
: Mr. Roos, the scheme you suggested is reasonable for relatively rigid milecules. But we're studying the non-rigid molecules with large-amplitude motions in the excited states and different equilibrium conformations in the ground and excited states. The simplest examples are formaldehyde (h2co) or acetaldehyde (ch3cho) molecules, carbonyl group of which becomes pyramidal in the lowest excited states. So, we should account for anharmonicity by studying some gradient ways on the excited state PES or, simpler, by studying one- and multidimensinal sections of the excited state PES by large-amplitude coordinates with optimization of rigid coordinates. Than we calculate anharmonic vibrational levels variationally and evaluate the Frank-Condon factors using our own programs similar to that of Laane.
: Currently such studies were performed on the CASSCF PESs using GAMESS(US) (see J. Struct. Chem., 42(2001), 66-75; IJQC 88(2002), 414-425; J. Mol. Struct. THEOCHEM 631(2003), 39-51) and for small molecules - on MR-CI and MR-AQCC PESs using COLUMBUS (see IJQC 96(2004),193-201; IJQC Vol 96(2004),589-597).
: It is doubtfully for us that two-dimensional section of the PES or the IRC curve could be obtained within numerical gradients in the form, useful for variational calculations.
: P.S. According to MOLPRO site, currently MOLPRO doesn't have an analytic CASPT2 gradient code, it is in development now.