[molpro-user] WORKAROUND: cannot get rohf to converge with larger basis sets

[Benj FitzPatrick]

Luis Pedro,
Thanks for the idea, it works quite nicely (I used a weighted average of the two
structures).

For the record, I tried rohf in g03 and it chokes completely.  Then I tried Cs
symmetry in molpro and the ground state, A", still has problems, but the first
excited state, A', converges fine.
Thanks again,
Benj 

---- Original message ----
>Date: Thu, 28 May 2009 11:51:48 +0100
>From: Luis Pedro Viegas <lpviegas at ci.uc.pt>  
>Subject: Re: [molpro-user] cannot get rohf to converge with larger basis sets  
>To: Benj FitzPatrick <benjfitz at uchicago.edu>
>
>Hi Benj
>
>"I repeated these calculations for the transition states leading away
>from this reactant and none of them had any problems."
>
>You could try the following:
>
>1-do an HF calculation for a geometry that you know that leads to
>convergence, and save the hf orbitals
>
>2-start the hf for this particular troublesome geometry reading the
>saved orbitals from the previous run
>
>
>Best regards,
>
>Luís Pedro Viegas
>
>
>On Wed, 2009-05-27 at 12:54 -0500, Benj FitzPatrick wrote:
>> Hello,
>> I'm using molpro 2008.1, patchlevel 31 and I'm having some problems with hf
>> converging using larger basis sets (aug-cc-pvdz and larger).  
>> 
>>  * The molecule is an open-shell doublet (input file and {locali,pipek;}
>> localized orbitals in molden format are attached), and this problem occurs for
>> both the cis and the trans conformers.
>> 
>> * 6-31G*, 6-311G**, and cc-pvdz converge, but adding diffuse functions or higher
>> angular momentum functions leads to non convergence.
>> 
>>  * I used mcscf to look at the first excited state and it is much higher in
>> energy (X = -191.420511770210, A = -191.295282002574) so there shouldn't be any
>> problems arising from a low-lying excited state.  Below is the ci vector output
>> as well showing X is pretty much dominated by one configuration.
>> 
>>  CI vector
>>  =========
>> 
>>  22222a000      0.9504701   0.0484423
>>  2222a2000     -0.0424724   0.7432273
>>  222220a00      0.0301000  -0.5814224
>>  2222a0200     -0.0020958  -0.2024822
>>  222202a00     -0.0073925   0.1907678
>>  2222aba00      0.1780930   0.0206321
>>  22220a200     -0.1285216  -0.0151333
>>  2222baa00     -0.1020188  -0.0095914
>>  22202a020     -0.0885933  -0.0053440
>>  2222aab00     -0.0760742  -0.0110408
>>  2220a2020      0.0030896  -0.0686470
>>  22022a002     -0.0611572  -0.0061246
>>  222020a20     -0.0029005   0.0538511
>> 
>> 
>>  * UHF has no problems converging (I even tried with aug-cc-pvtz).
>> 
>>  * I tried adjusting the level shifts, but I wasn't able to find a combination
>> that worked.  This was guess work on my part, so any tips for adjusting these
>> sanely would be great.
>> 
>>  * I repeated these calculations for the transition states leading away from
>> this reactant and none of them had any problems.
>> 
>>  * C1 symmetry (nosym) was used for all of the calculations.
>> 
>> Any help would be greatly appreciated.
>> Thanks,
>> Benj FitzPatrick
>> University of Chicago
>> _______________________________________________
>> Molpro-user mailing list
>> Molpro-user at molpro.net
>> http://www.molpro.net/mailman/listinfo/molpro-user
>