[molpro-user] Failure of convergence in CP-MCSCF NACM

[BELLSHAW Darren]

Dear mailing list,


I am using molpro for dynamics calculations of the CS2 molecule. However, these efforts fail when calculation the NACM elements between certain states (usually the 2nd and 3rd singlet state, but the precise point of failure seems to depend on the basis set/active space size) with an error message of the type:

"NO CONVERGENCE OF CP-MCSCF AFTER900 ITEATIONS AND 9 RESTARTS; FINAL ACTEST=  0.49D+00"

In an effort to isolate the problem, I've taken the geometry from the point where one of my dynamics calculations fails and tried to calculate the NACM elements in a standalone calculation with the following input:
__________________

gprint,basis,orbitals,civector
memory,100,m
symmetry,nosym;orient,noorient

angstrom
geometry={
 C      -0.025948935      0.001667053     -0.005782488
 S      -1.564330093      0.000265655      0.000791365
 S       1.574299060      0.000265547      0.000791127}

basis=tzvpp
{hf
wf,38,1,0}

{casscf
closed,11
occ,23
wf,38,1,0;state,4
CPMCSCF,NACM,1.1,3.1
CPMCSCF,NACM,1.1,4.1
CPMCSCF,NACM,2.1,3.1
CPMCSCF,NACM,2.1,4.1
CPMCSCF,NACM,3.1,4.1}
____________


The SA-CASSCF procedure converges smoothly, but the calculation again terminates with the same error as above when computing NACM elements. Is there anything obvious that may be causing this? Is there a way to increase the maximum number of iterations of the CP-MCSCF procedure? My active space is large - could this be a contributing factor? Many thanks in advance for any advice.


Best wishes,

Darren Bellshaw


Kirrander Group, University of Edinburgh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20160601/62388e84/attachment.html>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: not available
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20160601/62388e84/attachment.ksh>