BCCD(T) optimization

[Dave Moore]

Hi,

I am trying to re-run a slightly modified input file to do a BCCD(T)
optimization that ran fine under MOLPRO 2000.1.  The input file looks like:

*** START INPUT

basis = 6-311++G(3df, 3pd)

geometry = {} ! Cs symmetry

gthresh = dm, qm, sm

hf;
bccd(t), type=all; brueckner, 8000.2

optg, energy=1.0e-7, grad=1.0e-4
HESSIAN, NOMODEL
COORD, ZMAT

--- END INPUT


The HF and initial BCCD(T) calculations run fine, but immediately after the
BCCD(T) calculation .. before even the "end" of that program segment (which
I take to be the asterisk-row-delimited FILE/RECORD and CPU-TIME summary) ..
I get the following error message:

***** BEGIN ERROR SNIP *********

 Starting CP-QCISD iterations.     Author: G. Rauhut (1999)


 NOT ENOUGH MEMORY IN QCIDIP. ONLY   34 OUT OF   45 K-OPERATORS FIT

 ERROR EXIT

***** END ERROR SNIP ***********


I cannot see where I asked for any sort of "CP-QCISD" or "QCIDIP"
calculations ... (by analogy with MP2 calculations, I suppose they could be
part of some new analytical gradient method for CI/CC calculations).
Anyway, if I run a single point calculation commenting out the optg section,
everything works fine.  With MOLPRO 2000.1, a numerical gradient calculation
began after the end of the BCCD(T) section.  I thought maybe in 2002.3 I
needed to specify the NUMERICAL keyword explicitly, but that had no effect
on the outcome.

Any help on this matter would be greatly appreciated.  Thank you,

Dave Moore
FOM Rijnhuizen
Edisonbaan 14
3439 MN, Nieuwegein
The Netherlands
+31+30 609-6826
FAX:   603-1204