Dear AMBER users,
I recently tried to run AMBER22 Hamiltonian replica exchange simulations using
pmemd.cuda with Plumed implemented CVs for enhanced sampling. Upon examination
of the rem.log files, it seems to me that pmemd.cuda is not taking into account
the Plumed-contributed potential energy when it evaluates the success of the
individual trajectory exchange attempts across the replica ladder.
For example, if the two replicas differ only in terms of the Plumed contributed
potential energy, the exchanges succeeds in 100% of the time no matter how much
the potential energy differs. The expected behavior should be that the *sum* of
AMBER and PLUMED contributed potential energies is used to evaluate the exchange
attempts between the replicas. Right now, this evaluation is based solely on the
AMBER contribution.
I realize Plumed is not officially supported for Amber, so this bug report is
mainly intended to warn others about this problem.
Best,
Miroslav Krepl
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 17 2024 - 09:30:02 PDT
