[molpro-user] Question on F12 calculation for He atom

[Gerald Knizia]

On Thursday 27 August 2009 15:09, Yoshihiro Sumiyoshi wrote:
> I have been trying to calculate potential energy surfaces for complexes
> consisting of He atom at the UCCSD(T)-F12/aug-cc-pVQZ level of theory
> by Molpro 2009.1. Unfortunately, it does not work.
>
> In contrast, I could calculate successfully for complexes consisting of
> Ne atom with the same level of theory.
> I am not sure where the difference between He and Ne comes from.

Apparently there is no AVnZ/JKFIT basis set for the He atom. The paper where 
the other ones are from (PCCP 4, 4285 (2002), 
http://dx.doi.org/10.1039/b204199p ) does indeed not specify one. However, 
there are universal JKFITing sets available from the Turbomole library at 
http://www.ipc.uni-karlsruhe.de/tch/tch1/TBL/tbl.html (it can also output 
Molpro format), and the def2-QZVPP/JKFIT set from there would probably work 
for this purpose (but we didn't try!):

geometry={He}
basis={
   default,avqz
   set,jkfit
   ! He def2-QZVPP/JKFIT, F. Weigend, J. Comput. Chem. 29, 167  2008 .
   s He, 66.205029000,13.171713600,3.1622361000,0.81536220000
   c,1.1, 0.53033986000E-01
   c,2.2, 0.39465220220
   c,3.3, 0.91729877120
   c,4.4, 1.0000000000
   p He, 9.6128819820,3.2042939700,1.1643102000
   d He, 7.6739799000,0.98779470000
}

rhf
explicit,df_basis_exch=jkfit,ri_basis=jkfit
uccsd(t)-f12b,scale_trip=1

If this is too complicated or does not do the trick, the following should also 
work:

geometry={He}
basis=avqz
rhf
explicit,df_basis_exch=avqz/mp2fit,ri_basis=avqz/mp2fit
uccsd(t)-f12b,scale_trip=1

The MP2FIT sets are larger, however, and less well suited for Fock matrix 
fitting. For RI they may be better than the JKFIT sets.
-- 
Gerald Knizia