[gmx-users] Problems about the WALL used with NPT systems: possible bug?

[xu zhijun]

Dear GMx-users,
I have did lots of tests on a simple systems which only contain a water box and two 12-6 walls at z=0 and z=box. The simulations are performed in NPT with only semi-isotropic pressure coupling applied in z-direction. And of course the 2-d Ewald will be used in this simulation with ewald_geometry=3dc and PBC = XY.  However, in this simple system, the box length in z dimension keep increasing and results in a increased vacuum in the simulation box. Then by changing the wall-type to be 10-4, the pressure coupling works well with the vacuum disappearing (water density is also perfect). For your information, I have given the parameters I used in the simulation below. For 10-4 and 12-6 wall, in mdp file we only need to revise two parameters (wall_type and wall_density), as given below. I did this test in Gormacs 4.0/4.5/5.0. 

######################### ####################

 pbc             = xy

coulombtype     = PME
rcoulomb        = 1.3; 1.0
vdwtype         = switch
rvdw            = 1.2; 0.9
rvdw_switch     = 1.0; 0.8
ewald_geometry  = 3dc

tcoupl          = nose-hoover
tc_grps         = System
tau_t           = 0.1
ref_t           = 310
;pcoupl          = no
pcoupl          = Parrinello-Rahman 
pcoupltype      = semiisotropic
tau_p           = 15.0 15.0
ref_p           = 0    1.0
compressibility = 0    4.5e-5

nwall           = 2
wall_type       = 12-6;  for 12-6 wall, and "10-4" used for 10-4 wall
wall_r_linpot   = 0.0
wall_atomtype   = opls oplswall_density    = 114 114, only for "10-4" wall, for 12-6, no this term. 
wall_ewald_zfac = 3
 ###################################################################
If you have any suggestion about it, please let me know. Thanks a lot.
BestJerry