AMBER Archive (2004)

Subject: RE: AMBER: Turning off non-bond interactions

From: Chris Moth (chris.moth_at_vanderbilt.edu)
Date: Wed Apr 21 2004 - 10:00:14 CDT


>===== Original Message From david.evans_at_ulsop.ac.uk =====
 Physically, I can't see any reason why the atoms shouldn't be able to pass
through each other, but possibly there is some place in the code where 1/0 is
evaluated, or some other reason.
>
>So, is it possible to somehow turn off the VDW/electrostatic evalutions for
particular atoms or pairs of atoms entirely, without running the risk of this
behaviour?

Our Amber source code is off line at the moment - but I thought it might be OK
to toss out a 1/4 baked idea anyway...

AMBER must already deal with VDW radii of zero - these are in the force field
for hydroxyl hydrogens and protons in some water models (if memory serves).

So, I would look at the electrostatics term a bit more closely in the source.

It might be that charges of 0.0 are still being evaluated - which might lead
to math of (q1 x q2 = 0.0)/(very small inter atomic distance squared).

Perhaps there is an underflow in the evaluation of the denominator term.

A quick workaround for you might be to try extremely small vdw radii (.0001).
The repulsive term might keep your atoms far enough apart, and yet still give
you the simulation insight you are looking for.

I look forward to the final answer. Hope my idea does not lead you down too
wrong a path.

Chris

Chris Moth
chris.moth_at_vanderbilt.edu

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu