[gmx-users] Exploding temp/pressure.

[Nash, Anthony]

Hi all,

I¹m trying to equilibrate a Martini CG simulation from an initial
atomistic structure. Eq and Fc values were derived using an atomistic
system. I¹ve started the dt at 0.0005 for 600000 steps, moving through
0.001, 0.0015 and 0.002 for the same number of steps, using the .mdp
details below (with the exception of the time step and the output
delimiters - the remaining settings are from the Martini website). During
these steps the temperature dropped from approx 60000K down to 310K and
the potential energy dropped from positive figures down to around
-1.4*10^6. A final 600000 steps with dt = 0.002 showed a converged temp,
pressure, volume, potential energy (and total energy), and very little
change in the unit dimensions.

I now want to move the time step up to 0.025. I¹ve tried various dt values
(0.005, 0.01, 0.015Š 0.025) using the below setup, but each time dt causes
almost immediately structural explosion. The temperature and the pressure
rockets up. I changed the pressure coupling to berendsen and the same
happens. I have been through dozens of tau_p and tau_t combinations, each
one yields a trajectory of varying length before exploding (the protein
Œexplodes¹ at the exact time the the pressure and temperature increases).
I have also changed the thermostat to berendsen and Nose-Hoover, using a
range of 5*(dt*nsttcouple) and 20(dt*nsttcouple), respectively.


I would appreciate any thoughts or suggestions on how to solve my
temperature/pressure problem when I try increasing the dt to a CG value.

Thanks
Anthony


title                    = Martini
integrator               = md
dt                       = 0.025
nsteps                   = 300000
nstcomm                  = 100


nstxout                  = 10000
nstvout                  = 10000
nstfout                  = 0
nstlog                   = 1
nstenergy                = 1
nstxout-compressed       = 0
compressed-x-precision   = 0
;compressed-x-grps        =
energygrps               = collagen solvent
cutoff-scheme            = Verlet
nstlist                  = 20
ns_type                  = grid
pbc                      = xyz
verlet-buffer-tolerance  = 0.005
coulombtype              = PME ;reaction-field
rcoulomb                 = 1.1
fourierspacing       = 0.16 ;0.2  ;0.12
epsilon_r                = 15   ; 2.5 (with polarizable water)
epsilon_rf               = 0
vdw_type                 = cutoff
vdw-modifier             = Potential-shift-verlet
rvdw                     = 1.1

tcoupl                   = v-rescale ;berendsen ;v-rescale
tc-grps                  = collagen solvent
tau_t                    = 0.4 0.4 ;2.0 2.0
ref_t                    = 310 310

Pcoupl                   = parrinello-rahman
Pcoupltype               = semiisotropic
tau_p                    = 12.0  ;parrinello-rahman is more stable with
larger tau-p, DdJ, 20130422
compressibility          = 4.5e-5 0
ref_p                    = 1.0 1.0
refcoord_scaling         = com

gen_vel                  = no
gen_temp                 = 310
gen_seed                 = 473529
continuation = yes
constraints              = none
constraint_algorithm     = lincs
lincs-warnangle = 45
lincs-order=18
lincs-iter=4