[gmx-users] problem in extending a simulation run

[Mark Abraham]

Hi,

On Fri, Aug 18, 2017 at 8:41 AM manindersingh rajawat <
rajawat.manindersingh at gmail.com> wrote:

> Dear all,
>
> I want extend a 100ns run to 150 ns. For this I use following commands:
>
> gmx convert-tpr -s md_100ns.tpr -extend 50000 -o md_150ns.tpr
>
> gmx mdrun -v -deffnm md_150ns -c md_150ns.pdb
>
> But mdrun taken it as a new run


Sure. You implied that the checkpoint file was called md_150ns, but no such
file existed ;-)


> and began a full 150ns run:
> starting mdrun 'Protein in water'
> 75000000 steps, 150000.0 ps
>
> so i stopped it, and run with checkpoint input as following:
>
> gmx mdrun -v -s md_150ns.tpr -cpi state.cpt -c md_150ns.pdb
>
> Reading checkpoint file state.cpt generated: Fri Mar 24 05:39:47 2017
>
>
>   Version mismatch,
>     current program: VERSION 5.0.6
>     checkpoint file: VERSION 5.0.2
>

Generally you should continue simulations with the same version of GROMACS
you started, but there is unlikely to be a problem in this case.


>   Build time mismatch,
>     current program: Mon Jul 27 05:11:28 UTC 2015
>     checkpoint file: Fri Oct 24 17:49:05 UTC 2014
>
>   Build user mismatch,
>     current program: buildd at lgw01-56 [CMAKE]
>     checkpoint file: buildd at roseapple [CMAKE]
>
>   Build host mismatch,
>     current program: Linux 3.19.0-23-generic x86_64
>     checkpoint file: Linux 3.2.0-61-generic x86_64
>
> Gromacs patchlevel, binary or parallel settings differ from previous run.
> Continuation is exact, but not guaranteed to be binary identical.
>
> Using 1 MPI thread
> Using 4 OpenMP threads
> Compiled SIMD instructions: SSE2 (Gromacs could use AVX2_256 on this
> machine, which is better)
>

I would get GROMACS compiled properly for my expensive hardware so that it
runs more than twice as fast...


> WARNING: This run will generate roughly 19040 Mb of data
>
> starting mdrun 'Protein in water'
> 75000000 steps, 150000.0 ps (continuing from step 50000000, 100000.0 ps).
> step 50000600
>
> It starts well from the checkpoint, but due to little confusion i stopped
> it. and ran the following command:
>
> gmx mdrun -v -s md_150ns.tpr -cpi state_prev.cpt -c md_150ns.pdb
>
> It gives the following error
>
> Reading file md_150ns.tpr, VERSION 5.0.6 (single precision)
> Changing nstlist from 5 to 40, rlist from 1.2 to 1.223
>
>
> Reading checkpoint file state_prev.cpt generated: Fri Mar 24 05:36:56 2017
>
>
>   Version mismatch,
>     current program: VERSION 5.0.6
>     checkpoint file: VERSION 5.0.2
>
>   Build time mismatch,
>     current program: Mon Jul 27 05:11:28 UTC 2015
>     checkpoint file: Fri Oct 24 17:49:05 UTC 2014
>
>   Build user mismatch,
>     current program: buildd at lgw01-56 [CMAKE]
>     checkpoint file: buildd at roseapple [CMAKE]
>
>   Build host mismatch,
>     current program: Linux 3.19.0-23-generic x86_64
>     checkpoint file: Linux 3.2.0-61-generic x86_64
>
> Gromacs patchlevel, binary or parallel settings differ from previous run.
> Continuation is exact, but not guaranteed to be binary identical.
>
>
> -------------------------------------------------------
> Program gmx, VERSION 5.0.6
> Source code file:
> /build/gromacs-EDw7D5/gromacs-5.0.6/src/gromacs/gmxlib/checkpoint.c, line:
> 2204
>
> Fatal error:
> Failed to lock: md.log. Already running simulation?
> For more information and tips for troubleshooting, please check the GROMACS
> website at http://www.gromacs.org/Documentation/Errors
>
> When i run the previous command, that is
>
> gmx mdrun -v -s md_150ns.tpr -cpi state.cpt -c md_150ns.pdb
>
> It starts giving the same error to this also.
>
> Reading checkpoint file state.cpt generated: Fri Mar 24 05:39:47 2017
>
>
>   Version mismatch,
>     current program: VERSION 5.0.6
>     checkpoint file: VERSION 5.0.2
>
>   Build time mismatch,
>     current program: Mon Jul 27 05:11:28 UTC 2015
>     checkpoint file: Fri Oct 24 17:49:05 UTC 2014
>
>   Build user mismatch,
>     current program: buildd at lgw01-56 [CMAKE]
>     checkpoint file: buildd at roseapple [CMAKE]
>
>   Build host mismatch,
>     current program: Linux 3.19.0-23-generic x86_64
>     checkpoint file: Linux 3.2.0-61-generic x86_64
>
> Gromacs patchlevel, binary or parallel settings differ from previous run.
> Continuation is exact, but not guaranteed to be binary identical.
>
>
> -------------------------------------------------------
> Program gmx, VERSION 5.0.6
> Source code file:
> /build/gromacs-EDw7D5/gromacs-5.0.6/src/gromacs/gmxlib/checkpoint.c, line:
> 2204
>
> Fatal error:
> Failed to lock: md.log. Already running simulation?
> For more information and tips for troubleshooting, please check the GROMACS
> website at http://www.gromacs.org/Documentation/Errors
>
> Please help me to resolve it.
>

Don't have multiple simulations writing to md.log at the same time.

You'll also have fewer problems if you stop trying to manage your own file
names. If you want multiple parts that you have to manually recombine
later, use mdrun -noappend -s your.tpr -cpi. Otherwise mdrun -s your.tpr
-cpi and let the defaults do their job.

Mark