Re: AMBER: replica exchange out files continually overwritten

From: David Mobley <dmobley.gmail.com>
Date: Fri, 14 Dec 2007 11:18:21 -0800

All,

> Amber 10 will return to having complete output files for the entire run,
> rather than being overwritten. Amber9 does overwrite due to issues with how
> the
> actual exchanges are done in terms of file ownership among replicas.

Hmm, that sounds good.

> However, I have not seen a case with Amber9 where info files don't get
> written. Also, trajectory files should NOT be overwritten, just modified
> using
> an append. Are you sure the traj files only have 1 frame that keeps
> changing?

Sorry this wasn't clear. My trajectory files and rem.log are the ONLY
files that I get complete. Everything else gets overwritten or not
written. So yes, I am getting trajectory files. But no info files.

> is nwtx less than or equal to nstlim? is ntpr less than or equal to nstlim?
> both of these things work a little better in amber10, but for now make sure
> that you are actually telling sander to write at least once per exchange.

I have nstlim = 500, ntwx = 500, ntwe = 500.

I can send my input file again if that would be helpful. (In fact, I
can send a tarball of everything if you want to try reproducing it).

Thanks,
David


>
>
>
>
> On Dec 14, 2007 1:47 PM, David Mobley <dmobley.gmail.com> wrote:
> > All,
> >
> > I neglected to mention that my info files are not being written. And
> > everything else except rem.log and the trajectory files is being
> > overwritten every iteration.
> >
> > Thanks,
> > David
> >
> >
> > In fact, EVERYTHING except the
> >
> >
> >
> >
> > On Dec 14, 2007 9:17 AM, David A. Case <case.scripps.edu> wrote:
> > > On Fri, Dec 14, 2007, David Mobley wrote:
> > > >
> > > > I've used AMBER 9 quite a bit, but am just trying to run my first
> > > > replica exchange simulations. I appear to have MPI working properly
> > > > and the simulations are running just fine. This is on a xeon cluster.
> > > >
> > > > However, all of my output files for individual replicas (with the
> > > > exception of the trajectories) are apparently being overwritten every
> > > > exchange attempt. I briefly see contents in them in between exchange
> > > > attempts, for example:
> > >
> > > >
> > > > Energy files appear to only have one block of energies ever. However,
> > > > my rem.log file fills up just fine, as do my trajectory files.
> > >
> > > This is what you should expect. See the first paragraph in Section
> 6.9.2, p.
> > > 164 of the Users' Manual: the usual roles of mdout and mdinfo are
> switched,
> > > as explained there.
> > >
> > > Others with more Amber 9 remd experience can of course chime in here.
> > >
> > > ...dac
> > >
> > >
> > > -----------------------------------------------------------------------
> > > The AMBER Mail Reflector
> > > To post, send mail to amber.scripps.edu
> > > To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
> > >
> > -----------------------------------------------------------------------
> > The AMBER Mail Reflector
> > To post, send mail to amber.scripps.edu
> > To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
> >
>
>
>
> --
> ===================================================================
> Carlos L. Simmerling, Ph.D.
> Associate Professor Phone: (631) 632-1336
> Center for Structural Biology Fax: (631) 632-1555
> CMM Bldg, Room G80
> Stony Brook University E-mail: carlos.simmerling.gmail.com
> Stony Brook, NY 11794-5115 Web: http://comp.chem.sunysb.edu
> ===================================================================
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Dec 16 2007 - 06:07:35 PST
Custom Search