Error Exit - linearly dependent basis/illegal negative file offset

[Aiko Huckauf]

On Monday, August 26, 2002, at 05:43 PM, Anthony Scott wrote:

> However, it does seem more likely, in our experience and somewhat
> confirmed by the original post, to occur with augmented basis sets.

> One strategy may be to change the geometry ever so slightly so that
> the symmetry is broken (ie just past the symmetry cut-offs).  It is
> not nice, but seems to work.

> Of course this will not be a solution for optimizations but it may
> work for SPs.

The "BASIS LINEARLY DEPENDENT OR WRONG S" error occured for single
point calculations on the complex formed by propyleneimine and
hydrogen peroxide, a system that positively _always_ belongs to the
point group C1.

Indeed, the error never occured in any calculation with a non-
augmented basis set. When I ran MOLPRO in a test series, I used the
same complex geometry (defined in xyz format, not in z-matrix format)
with different levels of theory (MP2 and CCSD(T)) and basis sets
(cc-pVDZ and cc-pVTZ, aug-cc-pVDZ and aug-cc-pVQZ); the error
_exclusively_ occured with _some_ (!) of the augmented basis sets,
whereas MOLPRO completed the the other calculations without any
complaint:

MP2 cc-pVDZ          works fine
MP2 cc-pVTZ          works fine

MP2 aug-cc-pVDZ      works fine
MP2 aug-cc-pVTZ      ERROR: BASIS LINEARLY DEPENDENT OR WRONG S

CCSD(T) cc-pVDZ       works fine
CCSD(T) cc-pVTZ       works fine

CCSD(T) aug-cc-pVDZ   works fine
CCSD(T) aug-cc-pVTZ   ERROR: BASIS LINEARLY DEPENDENT OR WRONG S

This is what I do not understand. (I have to admit, though, that the
system under investigation is fairly large, so that the mentioned
level/basis combination might be a little bit too ambitious anyway.)

Aiko Huckauf

--
Dr. Aiko Huckauf                     Department of Chemistry
Fon: 780-492-4635                    University of Alberta
Fax: 780-492-8231                    Edmonton, Alberta  T6G 2G2
E-Mail: Aiko.Huckauf at ualberta.ca     Canada