AMBER Archive (1999)

Subject: (no subject)

From: ross_at_cgl.ucsf.EDU
Date: Thu Jul 22 1999 - 16:35:12 CDT

        I adjusted the density of the system by running constant volume MD and =
        decreasing the size of the periodic box gradually. However, upon =
        achieving a density of 1 and switching to constant pressure for the =
        remainder or the simulation, I encounter two problems.

The normal way to get to a density of 1 is run constant pressure.
At that point, some people like to switch to constant volume. The
stuff on equilibration in the Q&A on the amber web is worth a look
if you haven't seen it already.

        1. As the constant pressure MD proceeds, the pressure remains fairly =
        constant, but the density of the system is steadily decreasing, reaching
        ~.96 by 4 ps.

This seems like possibly an insignificant timescale to me, given
that big systems can easily take >50ps to even begin to equilibrate..
have you tried plotting energy components?

Perhaps in 'manually' compressing the system, you forced something
too quickly and it thus needs to relax for a time before reaching a
'comfortable' density of 1. Over what amount of time did you
reduce the volume in the 1st phase?

        2. The system encounters a fatal error involving SHAKE after 4.06 ps =
        stating that coordinate resetting cannot be accomplished, deviation is =
        too large.
        NITER, NIT, LL, I, and J are : 0 2 29 49 50=20
        However, UNLIKE the question in the commonly asked questions section of
        the AMBER web page, I have IFTRES=3D1 already since my system is very =
        large.

Better in my opinion to equilibrate at constant pressure
to avoid introducing strains that might be causing this.
You might also want to look at atoms 49 & 50 to see what's
going on in that region, and you could also try turning off
SHAKE for a bit to see if it will allow the system to relax
out of the situation.

        way around this error with constant pressure? I'm using 9A for the =
        nonbonded cuttoff (with PME), should I increase this value?

In my opinion, no.

Bill Ross