[molpro-user] Density Fitting

[H. -J. Werner]

Dear Glen,
the problem is that in benzene the localization is not unique. You can
use a method which eliminates the redundancy using

pipek,method=2

With diffuse basis sets you might want to use

pipek,method=2,delete=1 
or
pipek,method=2,delete=2 

This eliminates the contributions of the 1 or 2 most diffuse functions of each angular
momentum type to the localization criterion. For benzene and similar molecules
this gives better localization. 

Still the result of the localization is not unique. To remove the
ambiugity, use option mergedom=1.
See the manual for more details. A recent
application of all this is described in

J. G. Hill, J.A.  Platts, and H.-J. Werner,
Calculation of Intermolecular Interactions in the Benzene Dimer using Coupled-Cluster and
Local Electron Correlation Methods, Chem. Phys. Phys. Chem.  8, 4072 (2006).

Best regards
Joachim Werner


On So, 17 Sep 2006, Glen Jenness wrote:

>Dear Molpro users,
>I’m currently using the density fitting DFT-SAPT module in 2006v.1 of Molpro.
>I did a AVTZ (with VQZ and AVTZ as my coulomb (exchange) and MP2 aux.
>basis) with benzene-water.  When I did this, I noticed the following
>message in the benzene part of the calculation (it appears before the
>first iteration):
>
>NO CONVERGENCE in PIPEKS AFTER MAX ITERATIONS. LARGEST ROTATION ANGLE=
>0.00008498
>
>The run will complete, but I get nonsensical results for the interaction
>energy and components.  This problem also appears when I use the AVDZ
>basis (with VTZ and AVDZ aux.) with density fitting.
>However, this problem doesn't exist with AVDZ and VTZ (I haven't done AVTZ
>due to hard disk restrictions) without density fitting.
>I also have a second problem with AVTZ with density fitting (that's prob.
>due to a bad input, but I can't figure it out).  For some reason, it seems
>to act like density fitting isn't there with respect to disk space
>savings.  I did coronene-water with density fitting, and the disk used was
>~30GB, so for bezene water (with fewer functions) should take less disk
>space, but this isn't happening.
>If anyone can shed any light on either one of these, it would be greatly
>appreciated.  I've included the full outputs from both density fitting
>benzene-water runs.
>Thanks,
>Glen Jenness
>Jordan Group
>Department of Chemistry
>University of Pittsburgh
>
>

-- 
Prof. Hans-Joachim Werner
Institute for Theoretical Chemistry
University of Stuttgart
Pfaffenwaldring 55
D-70569 Stuttgart, Germany
Tel.: (0049) 711 / 685 64400
Fax.: (0049) 711 / 685 64442
e-mail: werner at theochem.uni-stuttgart.de