[lammps-users] About fix nve in DPD

Hi everyone,

I’m trying to use DPD to simulate a water-oil system. I followed a input script I found in the mail list archive. There is one line making me confused.

fix 1 all nve

According to Groot and Warren’s paper, DPD simulation is intrinsicly an NVT method. They did not mention anything about constant energy. But the beads won’t move at all without this fix!!
So how can I interprete this? Is DPD method a NVT or NVE in LAMMPS? or NVTE??

units lj
atom_style hybrid atomic bond
pair_style dpd 1.0 1.0 34387
communicate single vel yes
bond_style harmonic
read_data data.lammps


group water type 1
group decane type 2

define masses and interaction coefficient

#mass * 1
pair_coeff 1 1 131.5 4.5
pair_coeff 2 2 131.5 4.5
pair_coeff 1 2 161 4.5
bond_coeff 1 350 0.72

first equilibrate the initial condition

fix 1 all nve #???

create initial velocities

velocity all create 1.0 4928459 dist gaussian

change neighbor list parameters to avoid dangerous builds

neighbor 2.0 bin
neigh_modify delay 2

specify simulation parameters

timestep 0.04
thermo 100
thermo_style custom step pxx pyy pzz
variable itf equal (pzz-0.5*(pxx+pyy))*lz
fix 3 all ave/time 1 50000 50000 v_itf file itf.txt

dump 1 all atom 500 water.dump
dump_modify 1 scale no

run 200000

Thanks a lot


fix nve is the right way to go.

it is "constant" in energy in the sense that it doesn't
add or remove kinetic energy. the fix does not control
what other fixes or pair styles (in the case of dpd) are
doing, so it is the correct choice for your case.

i am constantly surprised by the number of people
that expect magic things from an nve ensemble (pun
is intented). guys, _please_ make some effort to have
at least some basic understanding in statistical mechanics
before asking questions about "ensembles and stuff".