[AMBER] SHAKEH hydrogen clusters

From: Pietro Aronica <pietroa.bii.a-star.edu.sg>
Date: Thu, 10 Sep 2020 20:19:39 +0800


I have an unusual system that I am trying to simulate with
pmemd.cuda_SPFP. The simulation fails at the start with the following error:

0: 19 - 20 and 4: 19 - 21
Hydrogen atom 19 appears to have multiple bonds to atoms 20 and 21 which
is illegal for SHAKEH.
Exiting due to the presence of inconsistent SHAKEH hydrogen clusters.

The thing is, 19 isn't a hydrogen: it's carbon, bound to hydrogens 20
and 21. Parmed confirms that the atom is in no way a hydrogen, either by
mass, atomic number, etc. Running the same simulation with sander
instead works.

I have tracked down similar answers in the mailing list, and it seems to
be due to a difference in computing language between C and Fortran that
changes the indices of the atoms. My system is unusual, but it should
work, as seen with sander.

Is there a way to fix the language difference? The previous responses
don't point to a solution.

Kind regards

Pietro Aronica

AMBER mailing list
Received on Thu Sep 10 2020 - 05:30:02 PDT
Custom Search