AMBER Archive (2003)

Subject: Re: AMBER: pmemd

From: james tomomi macdonald (j.macdonald_at_mail.cryst.bbk.ac.uk)
Date: Mon Aug 18 2003 - 14:52:32 CDT

Thanks,

I've reimaged with ptraj, its still giving high COM velocities but i
should be able to now complete the simulation.

I suspect there probably is something wrong with my run, but i can't think
what it may be. The COM velocities it is removing seem to be steadily
rising as the run progresses - it is now up to around 2.5ns.

I'll email the files you are interested in tomorrow.

cheers,

James.

On Mon, 18 Aug 2003, Robert Duke wrote:

> Date: Mon, 18 Aug 2003 12:51:16 -0400
> From: Robert Duke <rduke_at_email.unc.edu>
> Reply-To: amber_at_scripps.edu
> To: amber_at_scripps.edu
> Subject: Re: AMBER: pmemd
>
> James -
> Looks like possibly some higher energy stuff, but I would also note that it
> would be worthwhile trying to set iwrap=1. I quote from the amber 7 manual:
> "For very long runs, setting iwrap=1 may be required to keep the coordinate
> output from overflowing the trajectory file format". Using COM keeps the
> center of mass at the center of the box. In long runs (or probably shorter
> runs with higher velocities), you will still see the atoms spreading out in
> space unless wrapping is done. I would suspect a problem with the
> simulation unless it is very long, and actually, I am suspicious of the high
> COM velocity being removed (I typically see 0 here, but I spend more time
> writing code than doing simulations). I would be interested in a
> comparable sander 7 run, but with IDENTICAL run conditions, starting with a
> well minimized structure. The fact someone else is doing a run on another
> system with sander 7 without problems means nothing (we actually have a heck
> of a lot on nsec accumulated with pmemd running on big protein simulations
> with no problems in our group). And by the way, sander 7 may be broken
> under mpi for minimizations (problem recently reported to Prof. Case) - it
> gives different results fairly quickly when run with mpi as opposed to a
> uniprocessor version - so I currently don't trust sander 7 for mpi
> minimizations (I still use 6, or pmemd which agree in all implementations).
> Anyway, I would be interested in your starting files just out of
> minimization, the files right before the system goes south, and the "gone
> south" files. I also think it would be valuable for some of the real MD
> pros to speak up on this one. I am more of a software guy.
> Regards - Bob
>
> ----- Original Message -----
> From: "james tomomi macdonald" <j.macdonald_at_mail.cryst.bbk.ac.uk>
> To: <amber_at_scripps.edu>
> Sent: Monday, August 18, 2003 12:05 PM
> Subject: Re: AMBER: pmemd
>
>
> >
> > Hi Bob,
> >
> > Actually, having looked at it further, it appears that there has been
> > quite a high COM velocity during my simulation (which is being removed),
> > and some of the coordinates have gone over -999, so a couple of
> > lines read:
> >
> > -207.0612654-164.0880036-682.2872437-521.2708399-275.5480435************
> > -521.3560550-275.9826974************-521.9206418-275.9782509************
> >
> > for example this is a typical line from my output file:
> >
> > check COM velocity, temp: 0.018330 0.51(Removed)
> >
> > I haven't tried running it on sander 7, but i am thinking of trying to see
> > if i still get high COM velocities.
> >
> > I think the solution would be to subtract a constant value from all the
> > coordinates.
> >
> > A work mate has been running other simulations with almost identical input
> > parameters on a different protein on sander 7, but is not getting high COM
> > velocities.
> >
> > If you would still like to have a look at my files, then i would be happy
> > to email them to you.
> >
> > Thanks for your help,
> > Cheers,
> >
> > James.
> >
> >
> > On Mon, 18 Aug 2003, Robert Duke wrote:
> >
> > > Date: Mon, 18 Aug 2003 11:45:47 -0400
> > > From: Robert Duke <rduke_at_email.unc.edu>
> > > Reply-To: amber_at_scripps.edu
> > > To: amber_at_scripps.edu
> > > Subject: Re: AMBER: pmemd
> > >
> > > James -
> > > Have you tried to read it in either sander 6 or 7 (depending on which
> mode
> > > you are running pmemd in)? It could just be a corrupted file. If you
> wish,
> > > feel free to make a gzipped tarball of the various files needed to repro
> the
> > > problem, and I'll take a look. I would bet on a truncated or otherwise
> > > corrupted restrt, though (what happens if you redo the run that creates
> the
> > > restart - do you get exactly the same thing; is there any platform
> mixing
> > > with binary files?)
> > > Regards - Bob Duke
> > > ----- Original Message -----
> > > From: "james tomomi macdonald" <j.macdonald_at_mail.cryst.bbk.ac.uk>
> > > To: <amber_at_scripps.edu>
> > > Sent: Monday, August 18, 2003 11:18 AM
> > > Subject: AMBER: pmemd
> > >
> > >
> > > >
> > > > Hi,
> > > >
> > > > I am having a peculiar problem with pmemd. I have been able to run pm
> emd
> > > > with no problem for a few weeks, but i have recently got a problem i
> can't
> > > > understand. For some reason it has become unable to read my restart
> > > > coordinate file reporting the error:
> > > >
> > > > FATAL: Could not read coords from production_105.xyz
> > > >
> > > > I have looked at the file production_105.xyz manually (and it looked
> OK)
> > > > and tried regenerating it to see if there was some kind of one-off
> error
> > > > when it was written.
> > > >
> > > > I would be very grateful if anyone could help me, and i could forward
> on
> > > > the file should that be necessary.
> > > >
> > > > Thanks very much in advance.
> > > >
> > > > James.
> > > >
> > > >
> > > >
> > >
> > -----------------------------------------------------------------------
> > > > The AMBER Mail Reflector
> > > > To post, send mail to amber_at_scripps.edu
> > > > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> > > >
> > > >
> > >
> > >
> > >
> > > -----------------------------------------------------------------------
> > > The AMBER Mail Reflector
> > > To post, send mail to amber_at_scripps.edu
> > > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> > >
> >
> >
> > -----------------------------------------------------------------------
> > The AMBER Mail Reflector
> > To post, send mail to amber_at_scripps.edu
> > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> >
> >
>
>
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>

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