'Rasscf and CI energies differ'

Posted by Julien Lejeune on February 02, 2012 at 14:01:08:

Dear Molcas team and Molcas users,

I am currently facing some difficulties with the RASSCF module of the 7.6 version of Molcas. When trying to increase the size of my active space, I run into this kind of warning :
########################################################## ### Rasscf and CI energies differ.
### #######################################################

iteration 1
RASSCF energy -1947.5583301336
CI energy -1947.5582835561
relative difference 0.0000000239
Severe convergence problems. Maybe the active
space is unsuitable for this system?
=======================================================No convergence in the CI section: MAXJT will be increased to
26
etc.

To finally stop before any convergence of the wavefunction...

When I have a look at the shape of the orbitals after a few steps in the RASSCF procedure, everything looks fine, except the occupancy of the orbitals goes crazy (very different occupation numbers with respect to the previous size of the active space, while the orbitals I add are mostly unoccupied)...

I had no such kind of problem with identical systems in previous Molcas versions (7.0 and 7.2 namely).

Could you please tell me what are the origins of this divergence and how to limit it?

Thank you in advance,

Regards,

Julien Lejeune.

Follow Ups:

Post a Followup

Name:
E-Mail:

Subject:

if B is 1s22s22p1, what is Li?

Passfield:

Comments:
: Dear Molcas team and Molcas users, : I am currently facing some difficulties with the RASSCF module of the 7.6 version of Molcas. When trying to increase the size of my active space, I run into this kind of warning : : ########################################################## ### Rasscf and CI energies differ. : ### ####################################################### : iteration 1 : RASSCF energy -1947.5583301336 : CI energy -1947.5582835561 : relative difference 0.0000000239 : Severe convergence problems. Maybe the active : space is unsuitable for this system? : =======================================================No convergence in the CI section: MAXJT will be increased to : 26 : etc. : To finally stop before any convergence of the wavefunction... : When I have a look at the shape of the orbitals after a few steps in the RASSCF procedure, everything looks fine, except the occupancy of the orbitals goes crazy (very different occupation numbers with respect to the previous size of the active space, while the orbitals I add are mostly unoccupied)... : I had no such kind of problem with identical systems in previous Molcas versions (7.0 and 7.2 namely). : Could you please tell me what are the origins of this divergence and how to limit it? : Thank you in advance, : Regards, : Julien Lejeune.

[ Follow Ups ] [ Post Followup ] [ Molcas user's WWWBoard ]