[gmx-users] walls + domain decomposition

[Aranha, Michelle]

??Hello,

My system (3nmX3nmX14nm) ~ 17000 atoms ?consists of a nanotube membrane with water (tip4p) at both ends of the tubes. I require the use of 2d pbc and hence want to use walls at the simulation box boundaries. I'm using Gromacs (single precision) 4.6.1.  I initially equilibrate the system, make the system whole using trjconv, minimize it and then try to add 2 walls  (12-6) and start the run in 2d pbc. I have tried running in both threaded and mpi versions with 8 threads/ cores and I get segmentation faults for both with errors:
Error message: DD cell 0 0 0 could only obtain 136 of the 139 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.
Using just 2 cores and trying to simulate with only a box of TIP4P water and walls results in same error.
Simulation with just a box of  3-site SPC water gives the following error message:  DD cell 0 0 0 could only obtain 126 of the 130 atoms that are connected via constraints from the neighboring cells. This probably means your constraint lengths are too long compared to the domain decomposition cell size. Decrease the number of domain decomposition grid cells or lincs-order.
I tried decreasing lincs order, decreasing time step and I have also tried 10-4 type walls with no improvement.?

This is the walls segment of my .mdp file
.mdp file (Walls )

;Walls
nwall = 2
wall-atomtype = opls_113 opls_113
wall-type = 12-6
wall-r-linpot = 1 1
;wall-density
wall-ewald-zfac: 3
ewald-geometry = 3dc

Please advise.

Best regards,
Michelle