AMBER Archive (2009)

Subject: RE: [AMBER] PMEMD 10 overflow issues

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Sat Aug 22 2009 - 12:25:31 CDT


Hi Ben,

> I should have said that I am getting corrupted restart files for the
> explicit solvent simulations, due to what I think is overflow (******
> where values should be). Nothing unusual is happening to my structure,
> its definitely not blowing up. What I also dont understand is that the
> restart file should be the last set of coordinates from my trajectory,
> but my trajectory file does not appear to be corrupted.

This is indeed strange. Do you know which atoms are being given *'s. Is it
in the velocities or the coordinates? If it is the velocities which are
overflowing this would explain why it is not seen in the trajectory file but
this would be strange of itself since it would imply your system is blowing
up for some reason. I assume this takes a while for this to occur and it
would be strange for your system to explode after a significant amount of
simulation has been done.

It would be helpful to see exactly what you are simulating. What the systems
look like, what force field, what options and exactly at what point the
issue occurs. Without this info it is going to be fairly hard to debug. If
you are running explicit solvent with iwrap=1 and implicit solvent with
nscm>0 then you should not be ever running into this problem so please
double check that and also try visualizing it first, it there anything
obvious going wrong? If all looks good then I will see if I can reproduce
what you are seeing (if you send me your files etc) and we can debug the
problem further.

All the best
Ross

/\
\/
|\oss Walker

| Assistant Research Professor |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross_at_rosswalker.co.uk |
| http://www.rosswalker.co.uk | PGP Key available on request |

Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber