[gmx-users] About extend the run,,

[Justin A. Lemkul]

On 6/2/12 11:23 AM, rama david wrote:
> Thank you Justin for quick reply ....
>
> On Sat, Jun 2, 2012 at 8:12 PM, Justin A. Lemkul <jalemkul at vt.edu
> <mailto:jalemkul at vt.edu>> wrote:
>
>
>
>         To perform these task I am using following command
>         1. grompp -f <New mdp file just change in  saving output>  -t <.cpt>  -c
>         <.gro
>         file(gro file from position restrained run )>  -o new.tpr
>
>
>         2. tpbconv -s new.tpr -0 extend.tpr  -extend 30000
>
>         3. mdrun -v -deffnm extend  -cpi <.cpt> -append
>
>         Is these approach is correct??
>
>
>     No.  The new.tpr file contains instructions to run from 20 -> 30 ns.  There
>     is no need to then invoke tpbconv.  You would also need a new invocation of
>     mdrun, since the output frequencies won't match.  You don't want to append
>     those files with mdrun (nor will the checkpoint file likely let you)
>
>
> Extremely Sorry for my carelessness , New mdp files means the same as previous,
> with same number of steps, but just change in output frequency
>
> So as per your recommendation and experience  Is any alternative to do these
> from 20 ns???
>

You'll get a mismatch in your files (checkpoint, trajectory, energy) in terms of 
frame interval.  You should not try to append to these files or extend the run. 
  Just run a new simulation from 20 -> 30 ns.  The grompp command was correct, 
assuming you set "tinit = 20000" in the .mdp file and setting nsteps 
appropriately to give another 10 ns.  Then run it as a new simulation.  It will 
still be an extension of the first simulation (since the .cpt preserves the 
previous state), but without the mdrun -cpi -append mechanism, which in this 
case you don't want (or possibly can't use).

>         Second query;
>
>         To rerun the crash run, users give .cpt file as input to -mdrun,
>         My query is, There are two cpt file a) pre.cpt  b) .cpt ,
>         So which one has to given as input???
>
>
>     Use gmxcheck to understand the contents of each.  The timestamp will also
>     tell you a difference, as will reading mdrun -h for an explanation of what
>     the -cpt option is doing
>
>
> As per the link..http://www.gromacs.org/Documentation/File_Formats/Checkpoint_File
> I interpreted following things..
>
> 1. I have to use the state.cpt , but in the case  problem to these I have to use
> prev.cpt,  I can check there
>      content by gmxcheck..
>
>       But my Query is What are may be the potential problems ?? and how to find
> them ??
>
> Crash the run is very often with my system, that why I am worried..
>

If you're getting frequent crashes, you should investigate why this is happening 
rather than just plowing ahead.  Are there error messages in the .log or 
stdout/stderr output?  Are your energetic terms sensible?  If the system is 
crashing due to physical instability, you're wasting your time producing junk. 
If the crashes are occurring due to hardware or filesystem instability, that's 
something to take up with your sysadmin.

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================