[gmx-users] mdrun error: There is no domain decomposition for x nodes that is compatible with the given box

Seyed Mojtaba Rezaei Sani s.m.rezaeisani at gmail.com
Thu May 7 07:40:57 CEST 2015


Thank you Mark. As Justin suggested I increased the box size and now I can
run the job on 4 machines. The old md.log file can be found here:

https://www.dropbox.com/s/bbmqln6oqjyus0b/md.log?dl=0

My first choice for the box was a dodecahedron one with the option -d 1nm
which now is increased to 2 nm.
Thank you Justin for your useful help.

On Wed, May 6, 2015 at 4:10 PM, Justin Lemkul <jalemkul at vt.edu> wrote:

>
>
> On 5/6/15 7:26 AM, Mark Abraham wrote:
>
>> On Wed, May 6, 2015 at 9:01 AM Seyed Mojtaba Rezaei Sani <
>> s.m.rezaeisani at gmail.com> wrote:
>>
>>  Dear Justin
>>>
>>> Thanks for your kind response. I didn't notice the point you mentioned.
>>> It
>>> seems that grompp doesn't care the difference betwen vdwq and vdw-q.
>>> however, I corrected the script. The error remains even in the case of
>>> choosing 2 nodes. The only way to have a successful run is a serial
>>> calculation.
>>>
>>
>>
>> Then that's life (except that if your protein is non-spheroidal, you might
>> have better luck with the long axis aligned to the z-axis).
>>
>> The domain decomposition code expects bonded interactions to be of shorter
>> range than short-ranged non-bonded interactions, so will not produce a
>> decomposition that has a bonded interaction that starts in one domain,
>> crosses another domain, leaves by the opposite face, and reaches a third
>> domain. This is implemented by setting a minimum size for domains based on
>> many properties of the observed starting structure. There's a brief
>> summary
>> of these choices printed in every .log file.
>>
>> The use of couple-intramol triggers the generation of atypical
>> interactions, that are implemented like bonds as far as the DD code knows,
>> and for a protein those "bonds" are so much longer that they set a minimum
>> size that means a decomposition is infeasible.
>>
>> AFAIK, FEP on a whole protein isn't going to work at all anyway.
>>
>>
> True.  But is "DOX" isn't a protein; it's doxorubicin.  There is no reason
> that I can think of that would cause DD to fail unless the box is
> unreasonably small.  An extension of normal interactions out to 1.9 nm
> should be within normal limits if the box is of adequate size.  Perhaps the
> OP can give us more information (linking the .log rather than trying to
> attach it) and describing the box size.
>
>
> -Justin
>
> --
> ==================================================
>
> Justin A. Lemkul, Ph.D.
> Ruth L. Kirschstein NRSA Postdoctoral Fellow
>
> Department of Pharmaceutical Sciences
> School of Pharmacy
> Health Sciences Facility II, Room 629
> University of Maryland, Baltimore
> 20 Penn St.
> Baltimore, MD 21201
>
> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
> http://mackerell.umaryland.edu/~jalemkul
>
> ==================================================
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>



-- 
Seyed Mojtaba Rezaei Sani

Institute for Research in Fundamental Sciences (IPM)
School of Nano-Science
Shahid Farbin Alley
Shahid Lavasani st
P.O. Box 19395-5531
Tehran, Iran
Tel: +98 21 2310  (3069)


More information about the gromacs.org_gmx-users mailing list