Dear all,
Somebody has broken the storage of prmtop information in SANDER with one of the recent commits. The pointer m04 should point to atom names (set up in locmem.F90) but for some runs the address is wrong. I suspect something gets overwritten in memory. As a consequence most QM/MM tests fail (assignment of atomic numbers does not work).
This could have been prevented by simply running the tests before commiting changes to the git repository.
I am currently in the UK working on an implementation with Mark Williamson but instead of getting our work done we have to spend our time trying to figure out where and when the bug was introduced. I hope the person who is responsible for introducing the bug is going to fix this asap. We think this might be related to some commits by James Maier but are not sure right now. If anybody has an idea what happened or who introduced the bug, please let us know. Thanks!
All the best,
Andy
--
Dr. Andreas W. Goetz
San Diego Supercomputer Center
Tel : +1-858-822-4771
Email: agoetz.sdsc.edu
Web : www.awgoetz.de
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Feb 09 2012 - 09:30:03 PST
