[molpro-user] CI convergence

[jacek klos]

Hello,
Looks like your convergence is distance dependent, did you try to  
decrease threshold for
two el integrals, maybe that would help, the keyword is if I recall  
twoel=1.d-22 for example
in the gthresh line

Best
Jacek
On May 8, 2009, at 9:10 AM, S. Horvath wrote:

> Hello all,
>
> I am extremely new to Molpro, and as such I am trying to run a
> MR-SO-CISD calculation for the six lowest electronic states of the
> molecule IBr- using ECPs. I am having convergence issues with the
> Multi-reference internally contracted CI program (prior to any
> spin-orbit stuff), even though the error message sounds like it  
> pertains
> to the MCSCF program not being converged. Moreover, I can truncate the
> calculation to only run the MCSCF, and the calculation finishes, which
> leads me to believe that the problem lies with the MRCI program.
>
> Error:
> ?Error: MCSCF not converged. This error exit can be avoided using the
> IGNORE_ERROR option on the ORBITAL directive
>
> The error message then the "ignore-error option," but I don't really
> know how this would affect the results of my calculation if I ignore  
> the
> fact that the MRCI/MCSCF is not converged.
>
> I've looked through the manual and have not found anything helpful.  
> I've
> tried the "maxiter" keyword, but to no avail. I also read that the
> convergence issues arise because the active space is not chosen
> properly; however, I can run the same calculation at the equilibrium
> distance (~ 3.12ang) instead of at 12ang (which is what I want...),  
> and
> the calculation runs to completion. It sounds like the larger
> internuclear distance is adding an extra lay of difficulty to the
> calculation, but I'm not sure how or why.
>
> On a related note, does anyone know why for the calculation at the
> equilibrium, the SO energies of my 6 states are listed twice? Is  
> this a
> major problem, or just a silly bug when writing the energies to the
> output file? Just curious.
>
> Below is my input file for 12.0 ang. Any advice that you can provide
> would be most appreciated.
>
> Thanks so much,
> Samantha Horvath
>
> =============================
> ***,ibr-
> memory,500,M !4 GB memory
> gprint,basis,orbital,civector
> gthresh,orbital=1.d-6,civec=1.d-6
> molecule='IBr-'
>
> geometry={I;
>          Br,I,R}
> R=12.00 Ang
>
> basis={
> ecp,i,ECP46MDF;
> sp,i,ECP46MDF;c;
> s,i,0.030;
> p,i,0.023;
> d,i,0.3550,0.1851,0.1025;
> f,i,0.4330,0.2026;
>
> ecp,br,ECP28MDF;
> sp,br,ECP28MDF;c;
> s,br,0.041;
> p,br,0.030;
> d,br,0.7062,0.2639,0.1047;
> f,br,0.5515,0.2580;
> }
>
> cpp,init,2;
> I,2,1.028,,,1.247;
> Br,2,0.266,,,2.259;
>
> {hf;wf,15,1,1;orbital,2100.2}      !SCF to obtain sample wf. Used
> ground-state config.
>
> {mcscf;maxiter,50;failsafe;
> occ,5,4,4,0;
> closed,2,0,0,0;
> wf,15,1,1; state,2; weight,0.1,0.1;
> wf,15,2,1; state,2; weight,0.1,0.5;
> wf,15,3,1; state,2; weight,0.1,0.1;}
>
> {ci;maxiter,50;occ,5,4,4,0; closed,2,0,0,0; wf,15,1,1; state,2;
> weight,0.1,0.1; noexc; save,3010.1}
> {ci;maxiter,50;occ,5,4,4,0; closed,2,0,0,0; wf,15,2,1; state,2;
> weight,0.1,0.5; noexc; save,3030.1}
> {ci;maxiter,50;occ,5,4,4,0; closed,2,0,0,0; wf,15,3,1; state,2;
> weight,0.1,0.1; noexc; save,3050.1}
>
> {ci;maxiter,50;occ,5,4,4,0; closed,2,0,0,0; wf,15,1,1; state,2;
> weight,0.1,0.1; save,4010.1}
> {ci;maxiter,50;occ,5,4,4,0; closed,2,0,0,0; wf,15,2,1; state,2;
> weight,0.1,0.5; save,4030.1}
> {ci;maxiter,50;occ,5,4,4,0; closed,2,0,0,0; wf,15,3,1; state,2;
> weight,0.1,0.1; save,4050.1}
>
> lsint      !call spin-orbit code
>
> !so calc w/ mcscf
> {ci;maxiter,50;hlsmat,ecp,3010.1,3030.1,3050.1} !rerun CASSCF with  
> SO info
>
> !SO calc w/ mrci
> {ci;maxiter,50;hlsmat,ecp,4010.1,4030.1,4050.1} !rerun CI with SO info
>
> -- 
> Samantha Horvath
> McCoy Research Group
> 2144 Newman-Wolfrom
> *********************
> 120 W. 18th Ave.
> Box #234
> Columbus, OH 43210
> 614-688-8180
>
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user



__________________________________
Jacek Klos, Ph. D.
Department of Chemistry and Biochemistry
University of Maryland
College Park, MD 20742 USA
http://www.researcherid.com/rid/A-6457-2008
__________________________________





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20090509/3b8f4fe3/attachment.html>