RE: amber-developers: Block of ice

From: Yong Duan <>
Date: Mon, 11 Apr 2005 11:53:29 -0700

Thanks, Tom,

So, removal of COM motion (and rotation) should solve the problem. This
explains why we can go back.

-----Original Message-----
[] On Behalf Of Thomas E.
Cheatham, III
Sent: Monday, April 11, 2005 10:25 AM
Subject: Re: amber-developers: Block of ice

> I thought Berendsen's scaling makes "block of ice". But we have gone
> back to single-scaling (vs separate solvent/solute scaling) since
> AMBER 6/7? Any of you observed the "block of ice" thing again? Any
> explanation?

Single-scaling (or in fact *any* scaling; see Harvey et al), especially
if there is not -complete- energy conservation will over time lead to
this behavior. Sources of incomplete energy conservation in AMBER
include cutoff, constant pressure, SHAKE tolerances that are not very
low, time step, ... It is highly recommended that center of mass motion
be removed "often" where often is a debatable quantity ranging from at
every step to every half ns or so depending on how large an energy

For a greater discussion, see:

Harvey et al. JCC 19, 726-740 (1998) [better discussion/derivation] Chiu
et al. JCC 21, 121-131 (2000)

Received on Wed Apr 05 2006 - 23:49:58 PDT
Custom Search