[molpro-user] Implementing double hybrid density functionals

[Matthias Heger]

Hello,

I'm experimenting with implementing double hybrid density functionals (specifically, DSD-BLYP) in version 2015.1.25 and I'm running into some problems.

First, double hybrids seem to work only without density fitting. If I call df-ks and df-mp2, the MP2 part does not pick up the correct reference energy and scaling parameters from the preceding KS calculation.

Second, the parameters for the dispersion correction are apparently not implemented either, so I have to define them manually. From my attempts this seems to work only if I include dispcorr as a separate function call, not with a disp directive in the KS (or DF-KS) block. But then I'll have to add the dispersion energy manually, and thus can't use analytical gradients.


Also, DSD-BLYP includes scaled spin components from MP2, but DF-MP2 doesn't seem to use the scaled contributions even if I use scsgrd=1.
  The final energy is still the standard (unscaled) one.

Here's what I tried:

{df-ks,b,lyp
dh,0.70,0.44}
dispcorr,scale=0.37,alpha=60
df-mp2,scsgrd=1,scsfacs=0.909,scsfact=1.045


Is there a way to get around these problems?


Best,
Matthias


-- 
Dr. Matthias Heger
Department of Chemistry, University of Alberta
11227 Saskatchewan Drive NW
Edmonton, AB T6G 2G2, Canada

Office: CCIS 4-151
Lab: CCIS 4-108
Email: heger at ualberta.ca