From owner-nwchem-users  Tue May 13 11:54:30 2003
Received: (from majordom@localhost)
	by odyssey.emsl.pnl.gov (8.8.8+Sun/8.8.5) id LAA26758
	for nwchem-users-outgoing; Tue, 13 May 2003 11:40:37 -0700 (PDT)
Date: Tue, 13 May 2003 13:38:27 -0500 (CDT)
From: B Barnes <barnes@wuchem.wustl.edu>
Subject: Symmetry and ECP basis sets
To: nwchem-users@emsl.pnl.gov
Message-id: <Pine.GSO.4.05.10305131323400.13768-100000@wuchem.wustl.edu>
MIME-version: 1.0
Content-type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-nwchem-users@emsl.pnl.gov
Precedence: bulk

Hello all,

First a summary, then a difficult basis set question for developers at the
end.

A few days ago I wrote about problems with a benzenethiolate-gold
structure, and how enabling or disabling symmetry usage in NWChem made the
calculation converge to different electronic states and resulted in a
drastically different absolute electronic energy, and different behavior
in geometry optimizations.  The mail is archived here: 
http://www.emsl.pnl.gov:2080/docs/nwchem/nwchem-support/2003/05/0004.NWChem_4.1_symmetry_electronic_states

On the hunch of Ilias Sioutis (thanks!), I tried a different ECP basis set
for gold.  Apparently, the Stuttgart 1997 RSC ECP basis occassionally has
problems.  While I had previously tried (and had the same problem with)
LANL2DZ, the 'CRENBL' and 'CRENBS' basis sets of Ross, Ermler,
Christiansen, et al worked fine (produced well-behaved molecules with and
without symmetry).  Other groups have published work using the Stuttgart
and LANL2DZ basis sets for molecules of this sort, but typically using
DFT; the failure of those basis sets in our calculation was using MP2.

So my question is, and this is important for future calculations and even
as a footnote in a publication, *why* does this one particular ECP basis
set work where the others fail?  What is going on in NWChem such that it
makes a difference whether symmetry is enabled, and why does the CRENBL
basis set work better than the others?

Is it possible there is a bug in NWChem causing this behavior?  These
problems only occur in SCF/MP2, not DFT.  Example input file echoed below.

thanks,

Brian Barnes


Start  benz_Au_geom_Cs_3_2

echo


charge 0

geometry noautosym noautoz units angstrom
 C     -0.790404     -3.48425     0.00000
 H     -1.39723     -4.39504     0.00000
 C     -0.396098     -2.89186     -1.21149
 H     -0.696416     -3.33813     -2.16605
 C     0.380862     -1.72119     -1.21987
 H     0.690735     -1.27361     -2.17036
 C     0.780818     -1.12259     0.00000
 C     0.380862     -1.72119     1.21987
 H     0.690735     -1.27361     2.17036
 C     -0.396098     -2.89186     1.21149
 H     -0.696416     -3.33813     2.16605
 S     1.73384     0.361946     0.00000
 Au     -0.577377     1.75862     0.00000
end

ecce_print ecce.out

basis "ao basis" spherical print
  H library "cc-pVDZ"
  S library "cc-pVDZ"
  Au library "Stuttgart RSC 1997 ECP"
  C library "cc-pVDZ"
END
ECP
  Au library "Stuttgart RSC 1997 ECP"
END

set tolguess 1e-8

scf
  rhf
  nopen 0
  maxiter 100
  thresh 1e-8
end

mp2
  freeze core 8
end

task mp2 energy