[molpro-user] explicit fitting basis problem with -f12 methods

[Kirk Peterson]

Jason,

when you specify a default fitting basis set as you've done below on the ccsd(t)-f12 or mp2-f12 command line (or in the explicit directive), there are default contexts used, e.g., for df_basis the program looks for a vtz/mp2fit set below and the ri_basis would default to perhaps vtz/optri.  Neither of these are currently defined for the Be atom (or any of the alkali or alkaline earth metals I believe, this is a work in progress).  If you want to define your own, do it in the basis set block:

basis={

default,vtz     ! specify orbital basis

set,df
default,vtz/orbital

set,ri
default,vtz/orbital

set,hfdf
default,vtz/orbital

}

then:

ccsd(t)-f12,df_basis=df,ri_basis=ri,df_basis_exch=hfdf

These choices of fitting sets (i.e., just the orbital basis sets) will definitely give lousy results though. The 2009.2 version of Molpro has the def2-vnzpp Turbomole basis sets (these are not the correlation consistent vnz-pp basis sets) defined for all elements, both jkfit and mp2fit. If you don't have them, you can fetch them in Molpro format from the Turbomole basis set site: 
http://bases.turbo-forum.com/TURBOMOLE_BASISSET_LIBRARY/tbl.html
and insert them in your df, ri, etc. blocks like I did above (note that what Molpro refers to as mp2fit and jkfit are called cbasis and jkbasis in Turbomole).

regards,

Kirk


On Dec 4, 2009, at 1:56 PM, Jason Byrd wrote:

> Molpro users,
> 
> I am trying to do a proof of concept CCSD(T)-F12 calculation using the VTZ basis sets for the Be atom.  Reading through the mailing list, it seems that I can do this by defining 
> ccsd(t)-f12,df_basis=vtz,ri_basis=vtz,df_basis_exch=vtz
> However this does not work!  What I get is this:
>  
>  Cannot find default basis VTZ for atom Be
>  Type=FIT
>  Context=JKFIT
> 
>  Please specify a default basis or define basis sets for all atoms!
> 
> Is it not possible to define a df_basis_exch basis set explicitly?  My input is below and attached is my output file:
> 
> *** Be ccsd(t)-f12
> 
> memory,275,M
> 
> basis=vtz
> 
> geometry={
>         Be
> }
> 
> hf
> 
> ccsd(t)-f12,df_basis=avtz,ri_basis=vtz,df_basis_exch=vtz
> ---
> 
> Sincerely,
> Jason
> -- 
> Graduate Student
> Department of Physics
> University of Connecticut U-3046
> 2152 Hillside Road
> Storrs, CT 06269-3046
> 
> Office: P209 
> Tel: 860-486-3484   
> Email: byrd at phys.uconn.edu 
> <be-cont.out>_______________________________________________
> 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/20091205/4b6a6c06/attachment.html>