[molpro-user] FCI calculation for H3

[Peter Knowles]

This does indeed fail reproducibly. There is a bug which we will certainly try
to find and fix.

Note that the job does work if you use symmetry, which can be achieved by
giving a geometry either in internal coordinates or with the atoms placed in the
xy, xz or yz plane.

Peter

--
Prof. Peter J. Knowles
School of Chemistry, Cardiff University, Main Building, Park Place, Cardiff
CF10 3AT, UK
Telephone: +44 29208 79182 Fax: +44 2920874030
Email: KnowlesPJ at Cardiff.ac.uk Web:
http://www.cf.ac.uk/chemy/staff/knowles.html

>>> "Bastiaan J. Braams" <braams at mathcs.emory.edu> 31/08/2005 13:34:13 >>>
Dear Molpro user,

I would like advice about an FCI calculation for H3.  The following
job fails very quickly with a segmentation fault on my Intel system.

***,SP ENERGY FOR H3 doublet
memory,84,m
geomtyp=xyz
geometry
nosym
  3       h3d.xyz-00000
 -1.670901
H         1.78893575      1.73664337      0.41159189
H         2.32139768      1.27371181      0.47179363
H        -1.87184533     -1.76995798     -0.88345953
end
basis=avtz
rhf;wf,3,1,1
fci
---

(The specific geometry is immaterial.)  Perhaps the system is simply
too large for an FCI calculation, but there is something odd.  If I
change the basis to just vtz then the job runs fine in just a minute
or so.  I would have thought that if an avtz basis for H3 is too
large, then probably H3 on a vtz basis is a very large problem as well
and might take hours.  So I am puzzled that an FCI calculation for H3
using the vtz basis is successful in a minute or so, but the same
calculation on an avtz basis fails.  Any advice?

Bas
--
Bastiaan J. Braams
braams at mathcs.emory.edu 
Emory University, Atlanta, GA