Posted by Steven Vancoillie on November 05, 2014 at 10:10:43:
In Reply to: MOLCAS vs. similar programs for calculating non-adiabatic couplings/Stark shifts posted by Samuel Markson on September 05, 2014 at 23:43:42:
: Hi all--course-grained question here. I am interested in calculating the potential curves for alkali and alkali-halide diatomic systems (Rb2, NaI, and the like), for the ground and several (up through Rydberg) excited electronic states, including non-adiabatic couplings, along with dipole (and possibly multipolar) terms for these states. For context, I'm mainly an AMO physicist, usually content with performing TDSE calculations on curves produced by someone else, yet who is now compelled to enter a world where the basis sets aren't merely sets of orthogonal polynomials.
: I realize that this is, in principle, possible on many programs, but documentation doesn't always distinguish between what is possible and what is advisable. Can anyone fill me in on the strengths and weaknesses for something of the above nature, particularly of molcas vs. molpro? There _seems_ to be better documentation of excited state/non-adiabatic couplings calculations for molcas than for molpro, but as I don't have a molcas license (I'll note that I do have access to a molpro license), it's hard for me to judge first-hand whether it's wise to spring for a molcas license.
: All kibitzing is much appreciated.
Hi Samuel,
Molcas is certainly suited for this kind of thing. I don't think anyone can provide any real unbiased comparison, and I don't have so much experience with Molpro (if we used it, it was mainly for its CCSD(T) method). However, you could certainly benefit from having both and doing the comparison yourself. Luckily, there are now free binaries available for academic use here: http://www.kvant.kemi.uu.se/molcas/
I mainly used Molcas for casscf/caspt2 and rasscf/raspt2 for excited electronic states of large transition-metal systems. This method can certainly be used for what you want to do. However, with such small systems, you might go to even more accurate methods like MRCI or maybe CCSD(T) (but I don't know much about computing excited states with CCSD(T)).
grtz
Steven