AMBER Archive (2004)

Subject: RE: AMBER: MD simulation : problem

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Sat Jul 10 2004 - 00:35:14 CDT

Dear Sachin,

> While running the MD simulations on my system, I was
> getting an error message-
> NB pairs 397 8996882 exceeds capacity ( 9000000)....
> That means the density of the waterbox is increasing.
> So I increased the MAXPR parameter to 120,000.
> And by increasing the MAXPR, I have allowed the
> density to go as high as it can.

This is a very large number of pairs and is likely to lead to a very very
slow simulation.
You don't mention what the density of your waterbox starts at and what it
ends at. I assume you are running with periodic boundaries and constant
pressure. If this is the case the density will likely increase slightly as
the system equilibrates and will then stabilise. A plot of the density vs
simulation time would be useful to see if the density is reasonable.

> Now my problem is that, how does this increase in
> density affect my simulations. I am using a cuttoff of
> 999.0. And I have a system of 657 residues. Is ok to
> let the density to increase or to work at a lower
> cutoff?

Running a system of this size in solution with periodic boundaries and such
a big cutoff is crazy. You are going to get almost nothing done and you will
need hectares of ram. If you are using periodic boundaries the use of such a
large cutoff is unecessary since the dielectric properties of the water
shield the charges. A cutoff of 12 angstroms is a more suitable value. See
http://amber.scripps.edu/tutorial/polyA-polyT_New/minandmd3.html#5.1.2 for
an example of energy convergence and cpu time vs cut off size.

> Any help and comments would be highly appreciated !

Run with 12A cut off. Initially heat your system and let it equilibrate for
20 to 50ps at constant volume and then switch to constant pressure. Plot the
density, pressure, temperature energy etc as a function of time and see if
they converge to sensible values. E.g H2O ca 1g/cm^3.

All the best
Ross

/\
\/
|\oss Walker

| Department of Molecular Biology TPC15 |
| The Scripps Research Institute |
| Tel:- +1 858 784 8889 | EMail:- ross_at_rosswalker.co.uk |
| http://www.rosswalker.co.uk/ | PGP Key available on request |

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