Hi everyone:
This problem has come up repeatedly on the mailing list, and we have a
simple test case, but I don't see a simple solution.
Run the following with the attached files:
pmemd.cuda -i md.in -p cA2.prmtop -c restrt
You should get an inconsistent SHAKEH error, arising in gpu.cpp in the
gpu_shake_setup() routine. The code seems to think that atom 62 is a
hydrogen, but it is not.
Some notes:
1. The code might be counting atoms from 0, rather than 1, but that
would just make the reported error message misleading to users expecting
(as parmed does) to have atom numbers starting at 1. But there aren't
multiple bonds to hydrogens even if one interprets the atom numbers
reported as being 0-based.
2. This only seems to happen with prmtops created by parmed/chamber.
But the details of the atoms involved, including their masses, all look
OK.
3. If you figure it out, please add some comments to this routine to
make it easier to debug if a similar problem arises again.
...thx...dac
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
- text/plain attachment: md.in
- application/octet-stream attachment: restrt
Received on Tue May 22 2018 - 05:00:02 PDT
