For some time, I have avoided Z-matrix geometry optimisations in ADF.
However, with the latest ADF2002.02 I had high hopes for success.

Unfortunately, when attempting to optimise a molecule with planar
ligands (imidazoles) using internal coordinates, ADF fails. As far as I
can see, there are torsion angles which vary from -179 to +179 between
successive optimisation steps which is then reported as an angle step
size greater than the default of 10 and so around we go again. At the
same time, the gradients start going through the roof.

ADF2.3 was nice and stable and RELIABLE. You could shove just about
anything at it and it would normally converge. However, ADF2.3 does not
have all the new capabilities which are essential.

This is extremely annoying and a problem which we complained about
right back when ADF2000 first appeared and yet still seems to be
present. The inability to perform Z-matrix optimisations or linear
transits is a serious shortcoming and will eventually force me to
convert to another program system.

I am copying this to the members of adf-list in the hope that if we all
make enough noise, fixing the error will become the number one priority.
 I will attach an offending output to a subsequent mail to avoid
clogging up everyone's mail systems.

Signed Disgruntled in Warwick

Rob Deeth

Dr R. J. Deeth
Department of Chemistry
University of Warwick
Ph: +44 (0)24 76523187
Web: www.chem.warwick.ac.uk/iccg
Received on 2002-08-03 13:22:53