After a delay of only 30+ years(!), Amber finally has a minimizer that knows
how to work with SHAKE and SETTLE. I've uploaded code to the master branch at
gitlab: you need to use sander and the ntmin=3 option to access this.
Of course, most Amber users who carry out minimization just use it to get
started, and turn quickly to MD. But for people with more demanding needs,
the new option is available. It really became feasible (i.e. easy enough for
even someone like me to code it) based upon Jian Liu's "middle-scheme" MD
code. It is of special importance with rigid water models, since it is not so
uncommon for minimizations that don't use SETTLE to blow up, (and for
minimizations that do use SHAKE/SETTLE to get the dreaded LINMIN failure.)
Give it a try if you have a need for this, but be on the lookout for
problems. I've used this a fair amount, but it is still very new code, so
be careful. But one *can* get to RMS gradients of 10**-9 or even with bond
length constraints.
....dac
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Mon Sep 19 2022 - 15:00:02 PDT
