# AMBER Archive (2008)Subject: AMBER: RE: reference paper - pls help

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Fri Sep 05 2008 - 11:37:24 CDT

Hi Ruchi,

Such questions should really be posted to the amber mailing list which is
details.

I'm not sure you will find a paper discussing this issue, more likely a text
book since it really centers around the underlying theory for Molecular
dynamics. Look up the Virial theorem of Clausius which is used to calculate
the pressure. Essentially the pressure is:

P = (1/V)[NkT-(1/3)sum,sum(rij*fij)]

So at low temperature NkT is almost zero and so the pressure is effectively
minus the sum of over the product of forces and positions. Hence you can get
a very large negative pressure.

The barostat then does:

(dP(t)/dt) = (1/taup)(Pbath - P(t))

So this would give a very large correction due to the fact that P(t) is very
-ve. The barostat will attempt to correct for this and can massively
overcorrect leading to a rapid change in system volume which leads to system
instability.

As for saying that constant pressure with restraints can also cause problems
I believe that I am actually referring to the use of constraints in the form
of Belly. This can cause problems since there is a force on the frozen atoms
which is effectively zeroed which means that the virial is not strictly
correct and so the pressure and ultimately the box size will not be correct.
With actual restraints, that use a harmonic potential where the correct
derivative of the restraints is taken this should not be a problem since
then things are formally correct. You may have issues if you restrain your
system to a very poor initial structure or restrain large parts of your
system that span the entire unit cell, although you'd have to test this.

Also due to an issue with the way the coordinates are scaled during NPT
simulations you will, I believe, have problems if you don't update the refc
for each run. I.e. having -ref point to the original structure while running
multiple NPT runs can cause problems. The only work around I have found is
to reset -ref to the restart file read in as -c on each step, this is not
ideal but solves this problem.

Good luck

Ross

From: Ruchi Sachdeva [mailto:ruchi.imt_at_gmail.com]
Sent: Sunday, August 31, 2008 9:33 PM
To: ross_at_rosswalker.co.uk
Subject: reference paper - pls help

Dear Sir,

I have gone through one of your tutorials "Simulating a small fragment of
DNA", where you have mentioned that calculation of pressure is inaccurate at
low temperature and using constant pressure with restraints can also cause
problems. You have also mentioned this thing in AMBER mailing list in
context of a membrane protein. Sir can you please provide me the reference
where this obsevation has been reported so that I can clear my doubts such
as is this a general observation or it depends on some other factors also?
It will help me decide the correct ensemble to use during heating process.