On Wed, Apr 29, 2015, Jason Swails wrote:
>
> Note that some people use the Berendsen thermostat with a long coupling
> constant (e.g., tautp=10) as a way of doing a sort of "corrected" NVE
> simulation in which the velocity scaling is small enough to (hopefully)
> minimally impact the simulation while at the same time correcting for
> potential integration errors by bleeding the extra energy introduced by
> that (or adding energy lost by that).
Just a note to be careful here. The "floating ice cube" symptom can reappear
when you run long enough with ntt=1. (Previous "cures" for this, that involve
stricter tolerances on Ewald and shake, may no longer work when one is
approaching 10^9 time steps.)
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Apr 29 2015 - 09:00:03 PDT
