[gmx-users] temperature explosion in NVE ensemble

[Alex Robertson]

Hi
    I am trying to run a NVE ensemble featuring a short peptide in tip3p 
water.  No matter what I do my total energy always starts climbing rapidly 
until I end up with a lincs error. I have tried prequilibriating with P and 
T coupling on,only constraining hbonds, minimizing the energy in advance, 
shortening the timestep to .3 fs, using various radii for pme and cutoffs 
and increasing the lincs_order.  In all of these cases I still get all major 
components of my energy increasing rapidly.  I am just wondering if any more 
experienced users have resolved similar problems in gromacs.

Here are the relevant parts of my mdp:

dt = 0.002 ; This is the step I
nsteps = 2600000 ; total 5.2 ns
comm_mode = Linear
nstcomm = 1
nstxout = 500
nstvout = 100
nstfout = 0
nstlog = 100
nstenergy = 100
nstlist = 10 ; I have tried lowering this to 5
ns_type = grid
pbc = xyz  ; means yes
rlist = 2.2
rcoulomb = 2.5
coulombtype = Cut-off
epsilon_r = 1     ;  Could this value be the problem?
vdwtype = Cut-off ;
rvdw = 2.5
constraints = all-bonds
constraint_algorithm = lincs


Thanks so much,
Alex Robertson

PS
One thing I noticed is that in the GMX321 the manual recommends setting 
lincs_iter to 2 for NVE, but this option is not availible in GMX314.  Is 
there some way to get around this?


----------------------------------------------
  Alex Robertson

   Pande Lab
   Department of Chemistry
   Stanford University

650-387-4149
----------------------------------------------