Well, thinking about this some more, perhaps if you had a high-temperature
simulation there is a nontrivial possibility of getting a bad velocity in
there, especially if you had something like a Berendsen thermostat and a
gas or gas-like simulation, liquid with a vacuum bubble at the beginning.
So one in a thousand restarts could have this problem, perhaps, under
plausible circumstances. Sensible, then, to choose a lower scale given
the marginal value of added precision.
I agree, there's no sense in trying to change it, I just see things like
this and wonder what the rationale was. Especially when I'm pulling my
hair out wondering if mdgx is doing things right, and one of the metrics
is the continuity of restarts.
Dave
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Mon Nov 07 2011 - 15:00:03 PST
