[gmx-users] periodic surface minimization

[escajarro]

Dear list members,

I am trying to simulate an infinite layer of quartz in Gromacs. Due to my
needs (afterwards I need to place a large molecule on top of it) the surface
must be at least 12x12 nm^2. I generated the input files from a pdb file of
the structure, and adapted the OPLS force field to my current needs, adding
some compatible parameters from the literature. I am using the parameter 

periodic-molecules       = yes

in the mdp file, because if not I have a warning about inconsistent shifts. 

As Gromacs could not minimize the structure, I just took a smaller version
of the surface compatible with the cut-off radius, with the idea to minimize
it and then simply replicate it to obtain the large surface I need. Now
Gromacs was able to minimize the small system, but I realized that it was
not taking into account the bonded potentials across boundary conditions. I
did not manage to obtain a topology file by the use of pdb2gmx which
included the bonded potentials across periodic boundary conditions, so I
wrote my own program to generate them. I checked that the missing
interactions were properly added.

I run again the simulation with the right topology file for the small
system, and everything worked. The problem is that, when I replicate the
minimized structure in the x and z directions to obtain the structure I need
(and generating a new topology file with the right bond information,
changing the simulation box size properly, etc), Gromacs still can not
minimize the system, so that it does not 'explode' when I later try to
minimize the system after generating a velocity for every atom, even at very
low temperatures.

I also tried to minimize the original, large simulation box with a proper
topology file, to change the minimization method, the minimization step,
tried different constraints, etc, but without success. I have been working
on this problem for three weeks now and I run out of ideas, can any of you
help? I copy one example of mdp I used.

;constraints              = all-bonds
constraints              = none

; integration parameters
;integrator               = l-bfgs
integrator               = cg
;integrator               = steep
dt                       = 0.00
nsteps                   = 15000
periodic-molecules       = yes

; center of mass removal
nstcomm                  = 10
comm_mode                = Linear

; neighbour searching
nstlist                  = 10
ns_type                  = grid
pbc                      = xyz          ;periodic boundary conditions in all
directions
rlist                    = 1.5

; electrostatics
coulombtype              = PME
rcoulomb                 = 1.5

; van der Waals interactions
vdwtype                  = cut-off
rvdw                     = 1.5
DispCorr                 = EnerPres         ;tail corrections

; Ewald parameters
fourierspacing           = 0.12
fourier_nx               = 0
fourier_ny               = 0
fourier_nz               = 0
pme_order                = 4
ewald_rtol               = 1e-6
optimize_fft             = yes

; energy minimization
emtol                    = 10.0
emstep                   = 0.01

Thanks

--
View this message in context: http://gromacs.5086.x6.nabble.com/periodic-surface-minimization-tp5011423.html
Sent from the GROMACS Users Forum mailing list archive at Nabble.com.