[gmx-users] Periodic Molecules & Domain Decomposition Error

boags a. (ab24g12) ab24g12 at soton.ac.uk
Mon Feb 20 16:26:26 CET 2017


I've been attempting to simulate a 2D polymer in a system with a membrane in Gromacs 5.1.4 with the periodic molecules flag turned on. This runs fine on one core, but when I try to run in parallel the job hits the following error

Started mdrun on rank 0 Mon Feb 20 12:32:11 2017
           Step           Time         Lambda
              0        0.00000        0.00000

Not all bonded interactions have been properly assigned to the domain decomposition cells

A list of missing interactions:
         Proper Dih. of  38039 missing      1

Molecule type 'Strand'
the first 10 missing interactions, except for exclusions:
         Proper Dih. atoms 1357  678  685  669 global  1357   678   685   669

Program gmx mdrun, VERSION 5.1.4
Source code file: /home/ab24g12/Downloads/gromacs-5.1.4/src/gromacs/domdec/domdec_topology.cpp, line: 435

Fatal error:
1 of the 150450 bonded interactions could not be calculated because some atoms involved moved further apart than the multi-body cut-off distance (0.63 nm) or the two-body cut-off distance (1.26 nm), see option -rdd, for pairs and tabulated bonds also see option -ddcheck
For more information and tips for troubleshooting, please check the GROMACS
website at http://www.gromacs.org/Documentation/Errors

I'm aware that an error like this is normally due a system exploding, but the system, which is several hundred thousand atoms runs fine locally on one MPI thread and my GPU, whenever I increase the number of MPI threads, either locally or on a cluster the system reports errors such as the one above. I have tried changing the -rdd value to no success also.

Is domain decomposition throwing an error due to the periodic molecules flag being turned on when I run the system in parallel?

Any help would be greatly appreciated


More information about the gromacs.org_gmx-users mailing list