From:	peon[ AT ]medchem.dfh.dk (Per-Ola Norrby)
Date:	Tue, 28 Jan 1997 10:35:46 +0100
Subject:	Re: CCL:(2) basis set balance continued

Douglas A. Smith wrote:
>OK, so the real issue is BSSE and basis set flexibility for describing the
>space available to all electrons.  For the latter, would the use of CI or
>multi-reference-CI help alleviate the problem?  Would that be a more
>expensive cure than using a balanced basis?

        Don't!  Two reasons: first, adding correlation is usually more
expensive than adding basis, and second, it doesn't alleviate the problem!
Correlation tends to "spread the electrons" more efficiently among the
existing basis functions, but it cannot do anything about basis set
deficiencies.  In fact, correlated methods are MORE susceptible than HF to
deficiencies in the basis (see, for example, the article by Scheiner in
Rev. Comp. Chem., Vol 2, Chapter 5).  That's part of why you should have
H-polarization when doing MP2.  A caveat though: I don't know the situation
for DFT methods, I've got a feeling from my own work that, for example,
B3LYP results aren't very much improved when going from 6-31G* to 6-311G**.

>What, then, happens when one is performing calculations on systems
>containing both very heavy atoms and very light atoms, let's say, a
>lanthanide and hydrogen?  If one uses an all electron basis for both atoms,
>isn't the basis set unbalanced?

        You don't have to go that far.  Even LiF is supposed to be pretty
unbalanced with small basis sets.  On the other hand, I assume you don't
really need the absolute energy of LaH to several decimal places?  With
some experimentation, you should be able to find a combination of basis
sets that reproduces some significant experimental fact you're interested
in, then you can go on from there.  Since we are living in a virtual world
of finite basis sets, experimental verification is VERY important as soon
as you go into a new field.

>Is this at least part of the reason for
>effective core potentials (ECPs) -- in addition to the problem of just
>handling all the electrons in a reasonable time at reasonable cost?

        No, not really.  The ECP replaces the core, which doesn't
contribute significantly to BSSE.  There are other very good reasons for
using ECPs though, the savings and the cheap way to get SOME relativistic
effects for the core electrons (especially important in the 2nd and 3rd
transition metal row).

        Per-Ola Norrby


^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 *  Per-Ola Norrby, Associate Professor
 *  The Royal Danish School of Pharmacy, Dept. of Med. Chem.
 *  Universitetsparken 2, DK 2100 Copenhagen, Denmark
 *  tel. +45-35376777-506, +45-35370850    fax +45-35372209
 *  Internet: peon - at - medchem.dfh.dk, http://compchem.dfh.dk/




Similar Messages
01/27/1997:  (2) basis set balance continued
12/10/1995:  basis stes
05/01/1997:  calculation of chemical shifts (summary)
01/27/1997:  Re: CCL:basis set balance?
11/28/1995:  basis set mixing
09/12/1996:  Summary: ECPs
01/30/1997:  Mulliken population analysis
02/04/1997:  Mulliken Populations Summary
11/26/1995:  Combinning Different Basis Sets.
07/13/1998:  Summary : Mixing basis sets.


Raw Message Text