[gmx-users] Bond Decomposition Error

[Wenjuan Jiang]

Hello Tom,

I got something like decomposition errors when running in GPU clusters, and
later I found it is because of PBC unit cell angles. You can try to use
trajconv to get out of one pdb file from your simulation box. If the unit
cell is non-90degree, it will cause this domain decomposition errors.

Thanks,
Wenjuan

On Thu, Nov 8, 2018 at 10:44 AM tca1 <tca1 at rice.edu> wrote:

> Hi everyone,
>
> I've recently gotten a simulation of a 2D crystal surface set up in
> GROMACS using a combination of x2top and periodic-molecules. Things
> run okay for me when I do small-scale tests on a single core
> (minimization, NVT equilibration, etc), however when I try to scale
> this up and run it on multiple processors I get an error saying:
>
> Not all bonded interactions have been properly assigned to the domain
> decomposition cells
>
> This happens for any number of cores larger than 1, and it happens
> even during my minimization step. I don't believe this is the result
> of a poor equilibrium geometry, because when I run intermediate-sized
> tests on a single core, they finish correctly after only a handful of
> steps, even using fairly tight minimization settings. Additionally,
> the box itself is simple - just a cube with periodic boundary
> conditions applied - and I'm leaving out everything but the crystal
> surface in the simulation to reduce the complexity for debugging.
>
> I suspect the error has something to do with the periodic-molecules
> option and an unidentified oversight in my job setup, but I have no
> idea where to look. Does anyone have any suggestions?
>
> Thanks,
> Tom Allen
>
> --
> 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.
>


-- 
Best regards

Wenjuan Jiang