[molpro-user] questions about SCS-MP2

[Werner Győrffy]

Dear Li,

SCS-MP2 gradients in version 2012.1 are available only through DF-LMP2 
with full domain. An input for a geometry optimization would be 
something like this:

...
df-hf
df-lmp2,domsel=1,scsgrd=1
optg

I have recently made a patch to version 2012.1 including SCS-DF-MP2 
gradients not through the local MP2 program. In this case, an input for 
a SCS-MP2 geometry optimization looks like

...
df-hf
df-mp2,scsgrd=1
optg

To that, you need to download the 2012.1.nightly tarball from the Molpro 
website which is available to licensees.

Regards,

Werner Gyorffy.

On 01/08/2015 02:26 PM, Wanlu Li wrote:
> Dear Molpro users,
> I tried to run SCS-MP2 optimization calculation with the help of
> Molpro2012, but I met some problems when define the density fitting
> basis sets.
> The error is:
>
>
> /Using ITF DF-MP2 gradient implementation/
> /
> /
> / Number of core orbitals:          35 (  13   6  11   5 )/
> / Number of closed-shell orbitals:  22 (   8   4   7   3 )/
> / Number of external orbitals:     284 (  93  55  86  50 )/
> /
> /
> / Molecular orbitals read from record     2100.2  Type=RHF/CANONICAL
> (state 1.1)/
> /
> /
> / Cannot find default basis USERDEF for atom Au/
> / Type=FIT/
> / Context=JKFIT/
> /
> /
> / Please specify a default basis or define basis sets for all atoms!/
> /
> /
> Certain parts of my input file is:
>
> /basis={/
> /ecp,Au,ecp60mdf;/
> /spdfg,Au,ECP60MDF_AVTZ;c;/
> /ecp,I,ecp28mdf;/
> /spdf,I,AVTZ-PP;c;/
> /set,df;/
> /default,avtz-pp/mp2fit;/
> /}/
> /
> /
> /{rhf,maxit=100,orbprint=20;/
> /wf,nelec=114,sym=1,spin=0,charge=-1;/
> /!occ,9,2,2,0;/
> /}/
> /
> /
> /DF-MP2,SCSGRD=1,DF_BASIS=df,SCSFACS=1.2,SCSFACT=1/3;/
>
> Is there someone who knows how to input the density fitting basis or
> have some input files to run the SCS-MP2 calculations? Any suggestions
> would be greatly appreciated.
>
> regards,
> Li
>
>
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user
>