On Mon, Sep 26, 2016 at 10:57 AM, Hannes Loeffler <
Hannes.Loeffler.stfc.ac.uk> wrote:
> Hi,
>
> I get the following error with pmemd.cuda (coming from cuda/gpu.cpp):
>
> 1783: 3557 - 3558 and 1785: 3558 - 3559
> Hydrogen atom 3558 appears to have multiple bonds to atoms 3557 and
> 3559 which is illegal for SHAKEH. Exiting due to the presence of
> inconsistent SHAKEH hydrogen clusters.
>
> It is unclear to me what is happening here. The first water hydrogen
> of the system (index 3558) is bonded to the oxygen (index 3557) and the
> other hydrogen (index 3559). ParmEd does not report any other atoms
> bonded to index 3558. I do not know what 1783 and 1785 stand for.
>
> The prmtop file is a conversion (parmed chamber) from a CHARMM psf
> (using a CHARMM force field generated with CHARMM) file. I have other
> such setups which work just fine.
>
​Probably the naming issue reported previously:
http://archive.ambermd.org/201609/0274.html
Note that the latest version of ParmEd (the version on GitHub) has been
modified so that TIP3 and OH2 names for water residue and water oxygen,
respectively, is automatically changed to WAT and O (so that the SETTLE
assignment will work). I'm hesitant to release an update for AmberTools 16
as it changes behavior and would break people that have already changed
their input files to specify WATNAM and OWTNM for chamber-specific
topologies.
pmemd.cuda is hard-coded to support multiple constraints to H-atoms *only*
for the SETTLE routine (not SHAKE). So if TIPnP waters with H-H
constraints are not dispatched to SETTLE, you get this error message.
All the best,
Jason
--
Jason M. Swails
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Sep 26 2016 - 09:00:04 PDT
