[gmx-users] Replica Exchange Error

[Roland Schulz]

On Sat, Apr 11, 2009 at 10:52 PM, Joe Joe <ilchorny at gmail.com> wrote:
> Hi Justin,
> Thanks for helping me with this.
>
> On Sat, Apr 11, 2009 at 4:46 AM, Justin A. Lemkul <jalemkul at vt.edu> wrote:
>>
>>
>> Joe Joe wrote:
>>>
>>> How do I determine what those values should be? How does it affect the
>>> simulation?
>>>
>>
>> The header of your log file should have printed what the initial guesses
>> were. The -dd option sets how many PP nodes there are in the domain
>> decomposition; using 4 cores, there are probably 3 PP: 1 PME, so you
>> probably can't tweak this option, really.
>
> I do not get it. the input to -dd is a vector. What to the elements of the
> vector represent?

WIth the dd option you give the number of nodes in the 3 dimensions.
The product of the three number + the npme number has to be equal to
the total number. Look at the parallelization chapter of the manual.

>>
>> The -rcon option is explained in mdrun -h or the manual.  See what mdrun
>> guessed, then increase this value.
>
> I tried -rcon 3.0 and got the same error. The error only occurs when I do
> the exchange. If I increase the replex from 500 to 1000 then I do not see
> the error for 1000 steps. How is the replex associated with the domain
> decomposition?

It shouldn't be. It is probably only a statistically difference you
see. So if you let run longer with replex 1000 you will see it too.

Roland

>>
>> -Justin
>>
>>> thanks,
>>>
>>> Ilya
>>>
>>> On Fri, Apr 10, 2009 at 8:00 PM, Justin A. Lemkul <jalemkul at vt.edu
>>> <mailto:jalemkul at vt.edu>> wrote:
>>>
>>>
>>>
>>>    Joe Joe wrote:
>>>
>>>        Hi All,
>>>
>>>        I get the following error when running replica exchange. I have
>>>        32 replica's with each replica running on 4 processors (128
>>>        total). How do I decrease the domain decomposition cell size,
>>>        -dss? What should I set it to?
>>>
>>>
>>>    <snip>
>>>
>>>    You're getting some hints from mdrun, I'd start there:
>>>
>>>
>>>        -------------------------------------------------------
>>>        Program mdrun_mpi, VERSION 4.0.4
>>>        Source code file: domdec_con.c, line: 679
>>>
>>>        Fatal error:
>>>        DD cell 5 0 0 could only obtain 96 of the 108 atoms that are
>>>        connected via vsites from the neighboring cells. This probably
>>>        means your vsite lengths are too long compared to the domain
>>>        decomposition cell size. Decrease the number of domain
>>>        decomposition grid cells or use the -rcon option of mdrun.
>>>        -------------------------------------------------------
>>>
>>>
>>>    ...So manually set -dd and/or -rcon.
>>>
>>>    -Justin
>>>
>>>    --    ========================================
>>>
>>>    Justin A. Lemkul
>>>    Graduate Research Assistant
>>>    ICTAS Doctoral Scholar
>>>    Department of Biochemistry
>>>    Virginia Tech
>>>    Blacksburg, VA
>>>    jalemkul[at]vt.edu <http://vt.edu> | (540) 231-9080
>>>    http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin
>>>
>>>    ========================================
>>>    _______________________________________________
>>>    gmx-users mailing list    gmx-users at gromacs.org
>>>    <mailto:gmx-users at gromacs.org>
>>>    http://www.gromacs.org/mailman/listinfo/gmx-users
>>>    Please search the archive at http://www.gromacs.org/search before
>>>    posting!
>>>    Please don't post (un)subscribe requests to the list. Use the www
>>>    interface or send it to gmx-users-request at gromacs.org
>>>    <mailto:gmx-users-request at gromacs.org>.
>>>    Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>>
>>>
>>
>> --
>> ========================================
>>
>> Justin A. Lemkul
>> Graduate Research Assistant
>> ICTAS Doctoral Scholar
>> Department of Biochemistry
>> Virginia Tech
>> Blacksburg, VA
>> jalemkul[at]vt.edu | (540) 231-9080
>> http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin
>>
>> ========================================
>> _______________________________________________
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://www.gromacs.org/mailman/listinfo/gmx-users
>> Please search the archive at http://www.gromacs.org/search before posting!
>> Please don't post (un)subscribe requests to the list. Use the www
>> interface or send it to gmx-users-request at gromacs.org.
>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>
>
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>



-- 
ORNL/UT Center for Molecular Biophysics cmb.ornl.gov
865-241-1537, ORNL PO BOX 2008 MS6309