[molpro-user] CASPT2 error

Kirk Peterson kipeters at wsu.edu
Wed Jan 5 22:35:54 GMT 2011


Alexander,

13 iterations in the CASSCF is actually quite a bit for Molpro.  In any event it looks to me like the reference CI for the CASPT2 was very close to converging, so just do what it suggests and you should be fine.  Note that the option to increase is  "maxiti".  Try the following:

{rs2c,shift=0.3,thrvar=1.0d-8,thrden=1.0d-8;maxit,100,200;state,1,4}
{rs2c,shift=0.3,thrvar=1.0d-8,thrden=1.0d-8;maxit,100,200;state,1,5}

regards,

Kirk


On Jan 5, 2011, at 1:21 PM, Alexander Sokolov wrote:

> Dear MOLPRO Users/Developers,
> 
> I have a question regarding the following problem. I am using MOLPRO 2006.1 and I am running CASSCF(10x10)/CASPT2 calculations for the ground and excited states of a transition metal complex with ECP. Calculations for the GS and first two excited states complete without any problems. 
> However, when I'm running calculations for the 3d and 4th excited states, CASPT2 calculation gives a very strange error:
>  NO CONVERGENCE IN REFERENCE CI AFTER 50 ITERATIONS. DEMAX=  0.87D-07 VRMAX=  0.17D-07
>  PLEASE INCREASE MAXITI AND RETRY!
> 
> My CASSCF calculation converges in 13 iterations, so the convergence shouldn't be a problem.
> Here's the portion of my input file:
> 
> {rhf;occ,49;wf,98,1,0;maxit,100;orbital,2100.2}
> {multi,energy=1.0d-12,maxit=20;occ,54;closed,44;start,2100.2;wf,98,1,0;state,5;canonical,ci}
> {rs2c,shift=0.3,thrvar=1.0d-8,thrden=1.0d-8;maxit,100;state,1,4}
> {rs2c,shift=0.3,thrvar=1.0d-8,thrden=1.0d-8;maxit,100;state,1,5}
> 
> Does it mean anything? And is it possible to avoid this error?
> 
> Thank you very much for your help,
> 
> Alexander Sokolov
> 
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20110105/b9c26999/attachment.html>


More information about the Molpro-user mailing list