AMBER Archive (2004)

Subject: RE: AMBER: atom out of bounds (fwd)

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Thu Jul 22 2004 - 15:54:52 CDT

Dear Venkat,

> can somebody please explain what this error exactly means.

When sander starts a simulation it creates an initial virtual box around the
system. This is needed to calculate various properties, memory usage etc.
After a long simulation in which something diffuses a long way it can move
to the boundary of this box. At this point sander will halt with the above
error message. Restarting the simulation from the most recent restart file
will enlarge the size of this box and allow you to resume the simulation.

However, such errors are normally indicative of more serious problems with
the simulation. Normally a "blow up" causes this problem. In this situation
two atoms get too close together, either because they started too close or
in a very strained situation or the time step is set too long such that you
get integration errors. These three situations all lead to very large forces
that can cause your system to "explode". As soon as an atom gets to the edge
of the virtual box the program will halt.
 
> when I observed the final structure before the simulation stopped one
> of the n-terminus is extending almost perpendicularly (in my case i am
> expecting the n-terminus to move). If it is because of this n-terminus
> going out of the box can someone help me here.

Ok, so this is what casues the problem. The N-terminus must have diffused a
long way for this to happen. Are you sure you expect it to move this far. If
you look at the simulation over time does it look like reasonable motion? I
would take a good look at the various component energies vs time as well as
the temperature vs time. Plot these and look for any spikes and or excessive
values. These will be indicative of more fundamental problems with you
simulation.

> If i have to increase the box size please let me know how to do that.

If you believe the simulation is valid then you should just be able to use
the latest restrt file that was written, setting irest=1 and ntx=5 in your
input file. Restart the calculation and the box should automatically be
increased in size. Only continue the simulation, however, if you are sure
the results to date are valid.

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