[gmx-users] Thermostat for REMD simulations in implicit solvent

[César Ávila]

Dear all,
I am trying to set up a REMD simulation for a peptide (CHARMM ff) in
implicit solvent (OBC GB).
Following Bjelkmar et al* I am using stochastics dynamics integration with
an inverse friction constant of 91 ps-1, 5 fs timestep,
virtual-sites for hydrogens, all-vs-all nonbonded (no cutoff).  The full
.mdp file is attached at the end of this mail. The problem I am
facing is that after a while, the temperatures of all replicas start
dropping even below the lowest target temperature.
Would you suggest changing some parameter or the whole thermostat to prevent
this from happening?

@ s0 legend "Temperature"
    0.000000  310.811523
    5.000000  435.627045
   10.000000  417.161713
   15.000000  414.248901
   20.000000  399.390686
   25.000000  375.087219
   30.000000  338.131256
   35.000000  339.961151
   40.000000  319.424561
   45.000000  290.442322
   50.000000  289.587921
   55.000000  248.746246
   60.000000  253.192047
   65.000000  242.619476
   70.000000  256.051941
   75.000000  237.648468
   80.000000  231.938690
   85.000000  217.029953
   90.000000  211.447983
   95.000000  210.393890
  100.000000  208.518417
  105.000000  196.718445
  110.000000  219.245682
  115.000000  202.957993
  120.000000  193.128159
  125.000000  198.278198
  130.000000  175.304108
  135.000000  164.925613
  140.000000  195.024490
  145.000000  201.153046
  150.000000  211.160797
  155.000000  189.525085
  160.000000  191.156006
  165.000000  186.545242
  170.000000  186.885422
  175.000000  182.838486
  180.000000  174.960098
  185.000000  175.244049
  190.000000  179.517975
  195.000000  165.785416
  200.000000  189.871048
  205.000000  179.510178
  210.000000  152.527710
  215.000000  160.109955
  220.000000  163.564148









* "Implementation of the CHARMM ff in GROMACS" (2010) JCTC, 6, 459-466


; Run parameters
integrator          =  sd
dt                  =  0.005    ; ps !
nsteps              =  20000
nstcomm             = 1
comm_mode           = angular       ; non-periodic system

; Bond parameters
constraints             = all-bonds
constraint_algorithm    = lincs
lincs-iter          = 1
lincs-order         = 6

; required cutoffs for implicit
nstlist             =  0
ns_type             =  grid
rlist               =  0
rcoulomb            =  0
rvdw                =  0
epsilon_rf          =  0
rgbradii            =  0

; cutoffs required for qq and vdw
coulombtype         =  cut-off
vdwtype             =  cut-off

; temperature coupling
tcoupl              = v-rescale
tc-grps             = system
tau-t               = 91
ref-t               = 300

; Pressure coupling is off
Pcoupl              = no

; Periodic boundary conditions are off for implicit
pbc                 = no

; Settings for implicit solvent
implicit_solvent    = GBSA
gb_algorithm        = OBC
gb_epsilon_solvent  = 78.3
sa_surface_tension  = 2.25936

;Output control
nstxout             = 1000
nstfout             = 0
nstvout             = 0
nstxtcout           = 0
nstlog              = 1000
nstcalcenergy       = -1
nstenergy           = 1000

; GENERATE VELOCITIES FOR STARTUP RUN
gen_vel = yes
gen_temp = 300
gen_seed = 1993
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20101126/6f4a28fe/attachment.html>