Axel,
Yes, I have tried using time steps as small as 0.25fs. I am using 0.75fs because the original paper describing the potential uses 0.75fs. However, I am also having a similar problem using the press/berendsen thermostat with TIP3P potential described in the lammps manual. The simulation will start, but immediately the pressure becomes nan.
I have also tried to use press/berendsen and temp/rescale with the SPE/E benchmark provided on the LAMMPS website with no success. I have tried various damping parameters varying from 1-10000 on press/berendsen and 1-100 on temp/rescale (using fix nve for integration). Nothing produces a stable run for more than two or three time steps. Is there some other settings I need to specify? I can use the NPT thermostat but I am unable to get fluctuations of pressure down to an acceptable level.
I would appreciate any ideas, I have been stuck on this problem for a while.
Please have a look at my modified benchmark file for SPC/E water using temp/rescale and press/berendsen:
The thermo output only shows three steps before pressure runs away
Step Temp E_pair E_mol TotEng Press Volume
0 450.00625 -133281.51 0 -101089.66 -391.54182 357506.43
1 447.00613 -133281.51 0 -101304.28 975.97786 446564.48
2 444.066 -147173.36 0 -115406.45 nan nan
ERROR: Out of range atoms - cannot compute PPPM
units real
atom_style full
read_data data.spce
replicate 2 4 1
pair_style lj/cut/coul/long 9.8 9.8
kspace_style pppm 1.0e-4
pair_coeff 1 1 0.15535 3.166
pair_coeff * 2 0.0000 0.0000
bond_style harmonic
angle_style harmonic
dihedral_style none
improper_style none
bond_coeff 1 1000.00 1.000
angle_coeff 1 100.0 109.47
special_bonds lj/coul 0.0 0.0 0.5
neighbor 2.0 bin
neigh_modify every 1 delay 10 check yes
velocity all create 300 432567 dist uniform
fix 1 all shake 0.0001 20 0 b 1 a 1
fix 2 all press/berendsen iso 1.0 1.0 1000.0
fix 3 all temp/rescale 1 300.0 300.0 4.0 100.0
fix 4 all nve
timestep 2.0
thermo_style one
thermo 1
run 100