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

[Justin Lemkul]

On 5/5/15 8:48 AM, Seyed Mojtaba Rezaei Sani wrote:
> Dear gmx users,
>
> Perhaps my problem is a repetitive question in the forum but I couldn't
> find a trouble-shooter response for my case.
> I am running a stochastic dynamics calculation for a solution of
> doxorubicin. The following error occurs while the calculation runs:
>
> Fatal error:
> There is no domain decomposition for x nodes that is compatible with the
> given box and a minimum cell size of 1.9433 nm
> Change the number of nodes or mdrun option -rdd or -dds
>
> I have tried several numbers of nodes resulting in the same error. Here is
> the mdp file:
>
> integrator               = sd
> nsteps                   = 1000000
> dt                       = 0.002
> nstenergy                = 1000
> nstlog                   = 5000
> ; turn off trajectory writing
> nstxout                  = 0
> nstvout                  = 0
> ; cut-offs at 1nm
> rlist                    = 1.0
> nstlist                  = 10
> coulombtype              = pme
> vdw-type                 = cut-off
> rvdw                     = 1.0
> constraints              = all-bonds
> ; set temperature to 300K
> tcoupl                   = v-rescale
> tc-grps                  = system
> tau-t                    = 0.2
> ref-t                    = 300
> ; and pressure to 1 bar
> pcoupl                   = parrinello-rahman
> ref-p                    = 1
> compressibility          = 4.5e-5
> tau-p                    = 5
> ; and set the free energy parameters
> free-energy              = yes
> couple-moltype           = DOX
> init-lambda              = 0
> ; these 'soft-core' parameters make sure we never get overlapping
> ; charges as lambda goes to 0
> sc-power                 = 1
> sc-sigma                 = 0.3
> sc-alpha                 = 1.0
> ; we still want the molecule to interact with itself at lambda=0
> couple-intramol          = no

This is what triggers the error.  Note from the manual, for couple-intramol = no:

"All intra-molecular non-bonded interactions for moleculetype couple-moltype are 
replaced by exclusions and explicit pair interactions. In this manner the 
decoupled state of the molecule corresponds to the proper vacuum state without 
periodicity effects."

The added exclusions and pairs require larger DD cells to be constructed because 
these bonded interactions occur over larger distances than normal.  This limits 
the size of the DD cells and hence how many processors/cores you can use.  You 
haven't said how many you're trying to use, but the solution is simply to use fewer.

> couple-lambda1           = vdwq

The option is "vdw-q" but I don't know if grompp magically translates it.  Just 
check.

-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

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