Hello everyone,
I realize what happened. There are a 4 or 5 different places that the
header of the inpcrd file is read in (to allow for large atom counts). The
one place I missed I mistook for the 'old style' inpcrd file that we don't
use anymore. That is why it was omitted there.
To comment on the question "why doesn't it always overwrite temp0 even when
REMD is not there?" --- This is sander.MPI behavior, and it's documented.
TEMP0 is an input parameter, and it doesn't make sense to ignore this
input, IMO. In REMD restarts, the input file is not really ignored, since
it's assumed that the temp0's are scrambled in the input files, so it
presumably still uses all of the specified temperatures (it just prevents
you from having to re-arrange them yourself).
I'm preparing a patch and it should be ready to go shortly.
Thanks for catching this!
Jason
On Tue, May 1, 2012 at 5:39 PM, Niel Henriksen <niel.henriksen.utah.edu>wrote:
> I just started doing some REMD with pmemd from Amber12.
> I'm glad someone caught this.
>
> I had been using ascii restarts ... I wasn't sure about the status of
> netcdf restarts and they aren't mentioned in the manual.
>
> Here is what I found:
> - sander.MPI reads temps from the ascii restart correctly
> - pmemd.MPI DOES NOT read temps from ascii restarts (they come from mdin)
> - pmemd.MPI does read temps from netcdf restarts correctly
>
> Also, the reported bugfix DOES fix the problem. I don't really understand
> the history behind all the inpcrd formats being checked in inpcrd_dat.F90
> but it must not be doing what we think it is.
>
> Maybe someone else can comment from here.
>
> Bottom line: BEWARE if using ascii restarts and REMD in pmemd.MPI
>
> --Niel
>
> ________________________________________
> From: case [case.biomaps.rutgers.edu]
> Sent: Tuesday, May 01, 2012 10:01 AM
> To: amber-developers.ambermd.org
> Subject: [AMBER-Developers] [pren.mail.utexas.edu: FW: REMD with AMOEBA]
>
> The message below seems to relate to REMD inside pmemd. Any comments from
> those involved with this? (Seems to relate to sander6 or pmemd 2.01(?)
> formats.) I'm confused by the subject line, since I don't think(?)
> there is support for REMD in pmemd.amoeba.
>
> ...thx...dac
>
> ----- Forwarded message from "Ren, Pengyu" <pren.mail.utexas.edu> -----
>
> From: "Ren, Pengyu" <pren.mail.utexas.edu>
> To: case <case.biomaps.rutgers.edu>
> Subject: FW: REMD with AMOEBA
> Date: Tue, 1 May 2012 15:03:27 +0000
>
> Robert Best suggests there is a bug in the (regular and AMOEBA) pmemd
> inpcrd_dat.F90 which causes the temperature not to be read from the
> restart file. It is the line 141 in the attached file. The addition of
> "local_temp0" should fix this problem.
>
>
> From: Robert Best <rbb24.cam.ac.uk<mailto:rbb24.cam.ac.uk>>
> Date: Friday, April 27, 2012 2:54 AM
> To: Pengyu Ren <pren.mail.utexas.edu<mailto:pren.mail.utexas.edu>>
> Subject: Re: REMD with AMOEBA
>
> In the amber development version it is just line 141 in inpcrd_dat.F90.
> I've attached the corrected file.
>
> I assume everyone uses binary or netcdf restarts now so that's why it
> wasn't picked up?
>
> Rob
>
> Robert Best
> Royal Society University Research Fellow
> Department of Chemistry
> Lensfield Road, Cambridge CB2 1EW
> http://www.ch.cam.ac.uk/staff/rbb.html
> Tel: +44-(0)1223-336470 | Fax: +44-(0)1223-336362
>
>
>
> >
> > On 4/26/12 10:21 AM, "Robert Best" <rbb24.cam.ac.uk<mailto:
> rbb24.cam.ac.uk>> wrote:
> >
> >>
> >> I've uploaded a new pmemd with some bugfixes relating to reading
> >> formatted restart files with REMD. Actually, this bug was present in the
> >> standard PMEMD and I had just copied it over. The bug caused the
> >> temperature not to be read from the restart file.
> >>
> >> Another thing I noticed in your prmtop files is that in the
> >> ATOMS_PER_MOLECULE section, the first hydrogen of ALA5 has been split
> >> into a different molecule. I don't think this has any real effect on the
> >> simulation, only the way that iwrap works. This was also causing
> problems
> >> with the restarts, but is easy to fix by hand.
> >>
> >> Robert
> >>
>
> ----- End forwarded message -----
>
> --
>
> ================================================================
> David A. Case | case.biomaps.rutgers.edu
> BioMaPS Institute and Dept. of |
> Chemistry & Chemical Biology | fax: +1-732-445-4320
> Rutgers University | office: +1-848-445-5885
> 174 Frelinghuysen Road, Rm. 208b | cell: +1-609-651-6817
> Piscataway, NJ 08854 USA | http://casegroup.rutgers.edu
> ================================================================
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue May 01 2012 - 16:00:03 PDT
