[molpro-user] midbond functions and F12 methods

[Grant Hill]

Dear Jessie,

What happens if you just try this? I.e., place a dummy atom (I suppose the choice of atom to dummy will have some affect as the basis functions will be different) somewhere in the bond. Does the energy change in a sensible way compared to the same system without the mid-bond dummy? Counterpoise corrected F12 calculations seem to work fine and what you want to do seems superficially similar.


Grant


On 2 Mar 2012, at 15:26, LUTZ J. wrote:

> 
> Greetings,
> 
> I'm curious how to define midbond functions in the context of explicitly correlated calculations. When defining midbond functions with conventional methods, a dummy atom is placed at a particular point in space and then the midbond functions are assigned as belonging to the ghost atom in the basis set definitions. Many people have had success running calculations in this way. However, when using the explicitly correlated coupled cluster methods and explicitly defining basis functions, one must define jkfit, mp2fit, and optri basis sets corresponding to the ghost atom as well. I know this has been done before, as in say J. Chem. Phys. 136, 074301 (2012), but after contacting that paper's author, it seems that his input file worked in the 2009 version of Molpro, but not in the latest version. Since the above paper showed that midbond functions could indeed improve the results of explicitly correlated calculations, I think it is worth clearly explaining how to use them in the latest version of Molpro.
> 
> My best guess was to simply copy the midbond basis functions explicitly defined for the standard basis set to each entry for the density fitting sets?
>  
> Thanks for your time,
> 
> Dr. Jesse James Lutz
> Postdoctoral Research Fellow
> Durham University
> United Kingdom
> 
> 
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user