Re: CCL:DNP in DMol <--> 6-31G** in Gaussian?

[Damian Gregory Allis <damian(at)somewhereville.com>]

 Greetings, Nicolas
	I've had quite a bit of experience using DMol3 (at NCSA) and GGA
 functionals for calculating inelastic neutron scattering spectra and
 examining the deformation of rigid molecules in the solid-state, which
 also means doing side-by-side DMol3 and Gaussian/GAMESS comparisons of
 isolated molecules for structural comparisons.
	Generally speaking, dnp and 6-31G** (as reported in the Delley papers
 as well) line up the best from among all available choices.  That said,
 I've not had an experience yet where the DMol3 geometries were closer
 to crystallographic distances than the Gaussian/GAMESS jobs.  The
 standard molecular treatments of the "familiar" atom-centered
 QC
 packages simply do a better job of getting the structure right in the 7
 or so systems I've worked on, so be less concerned with the DMol v.s.
 Gaussian/GAMESS structures and more concerned with those same
 comparisons to any structural data you have.  The DMol3 normal mode
 calculations, however, do a phenomenal job of reproducing the
 vibrational spectra and intensities in the INS work despite having
 slightly worse geometries.
 For a useful geometry comparison, I direct you to the pagodane paper
 below.  The BH4 paper below is biased towards frequencies only but
 provides another useful comparison.  Unfortunately, I've not played
 around much with the LDA functionals, so can't provide any useful info
 about differences you'll see there.
Allis D.G., Prinzbach H., and Hudson B.S. "Inelastic Neutron
 Scattering
 Spectra of Pagodane: Experiment and DFT Calculations." Chemical
 Physics
 Letters, 386 (2004) 356-363.
Allis D.G. and Hudson B.S. "Inelastic Neutron Scattering Spectra
 of
 NaBH4 and KBH4: Reproduction of Anion Mode Shifts via Periodic
 DFT."
 Chemical Physics Letters, 385 (2004) 166-172.
 cheers,
 Damian
 On Apr 19, 2004, at 9:28 AM, Nicolas Saettel wrote:
>   Hi all,
>  I would like to reproduce LDA calculations that were done using the
>  double numerical basis set with polarization functions in DMol (DNP).
>  Am I correct in using SVWN/6-31G** in Gaussian?
>   Thank you in advance
>   Nicolas Saettel
>  --
>   Nicolas Saettel, Ph.D. http://nicolas.saettel.free.fr
>   Laboratoire de Chimie des Interactions Moliculaires
>   CNRS UPR 285 - Collhge de France
>  http://www.college-de-france.fr/chaires/chaire10/index.htm
>  11, Place
>  Marcelin Berthelot - 75005 PARIS
>   Tel.: 01.44.27.13.63 FAX.: 01.44.27.13.56
 #8^(=)+<
 p.s. Damian apologizes for the delay.
   Locale : www.somewhereville.com
 Delivery : damian(at)somewhereville.com
  AOL IM  : swville
 "You don't need money to be rich anyhow.
 Spending yourself is what it's all about."
 -Marillion, "Rich"