Dear all,
After a long debugging session I found a bug in AMBER's semiempirical integral code that was quite hard to track down.
This bug affects all NDDO based QM calculations with atoms that are separated by 10 Angstrom or more. Without going into more detail, the bug is not always triggered. It depends on the specific geometry and atom ordering and is more likely to make trouble if d orbitals are present.
A bugfix will be availabe, soon. It should fix all SCF convergence problems we have seen as of late.
Thanks to everybody for their help along the way, reporting problems with semiempirical QM/MM simulations and providing structures for testing/debugging!
All the best,
Andy
--
Dr. Andreas W. Goetz
Assistant Project Scientist
San Diego Supercomputer Center
Tel : +1-858-822-4771
Email: agoetz.sdsc.edu
Web : www.awgoetz.de
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Nov 16 2012 - 17:30:02 PST
