Molcas Forum

SebastianS

Member

From: San-Sebastian (Spain

Registered: 2018-03-15

Posts: 2

Hints on doing state-specific CASSCF dissociation curves

Dear Molcas Users,

I am interested in a dissociation profile of the homonuclear diatomic molecule of one particular electronic state (for equilibrium geometry, it is the ground state). In fact I am going to analyze the topology of the density of the system along the dissociation, and that is why I am interested in the fully state-specific CASSCF computations, not the state average ones, since the topology of density is sensitive enough to distinguish the differences...

Do you have any hints on how to follow effectively one electronic state along the dissociation in order to obtain smooth dissociation profiles and avoid variational collapse to other state with much lower energy, using SS-CASSCF?

Of course, along the scan on internuclear distances, I am doing small steps in the displacements (circa 0.01A), and I am using converged NOs from the previous step as a guess to current SS-CASSCF. I also take advantage of ciselect and give as input CI coefficients for the desired root from previous step as well, but it has it limits.

Can anything else be done on that matter? Like maybe some particular SCF converging algorithm works better for such tasks, some criteria for orbital rotations etc?
Or maybe it requires completely different approach?

Best Regards,
Sebastian

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Hints on doing state-specific CASSCF dissociation curves

I would start from a SA-CASSCF calculation, and maybe try to get SS-CASSCF results from there. I'm not sure I understand/agree with your reason for requiring SS calculations. You say the topology of the electron density is sensitive to the difference between SS and SA, and that's probably true, but will it be a significant difference? Will it be more important than the instabilities that you'll probably get with a SS approach? Note that SS or SA refers only to how the orbitals in which the CASSCF wfn is expressed are optimized. In the Full-CI limit you'd get exactly the same result. Also, from a SA-CASSCF calculation you still get state-specific natural orbitals and densities.