Dilute coarse-grained polymer simulation in implicit solvent using dpd/tstat pair style: diffusion dependence of friction coeff

Hi all,

I am attempting to simulate dilute concentrations of a coarse-grained polymer in implicit solvent.

I was previously using Langevin dynamics but in order to simulate hydrodynamic interactions I have instead been attempting to use dissipative particle dynamics.

Using lj96/cut and dpd/tstat for the pair style (using hybrid/overlay) I have been running simulations but there is no dependence of the diffusion coefficient on the friction coefficient.
I have tried a range of coefficients (values for gamma) from 0.0002 to 20. I was hoping to scale the friction coefficient in order to match diffusion in atomistic simulations with explicit solvent.

Each monomer unit is made up of 3 coarse-grained site types and I have 120 particles in my system of box length ~62 Angstroms. Using lammps 2nd April 2011.

Any suggestions would be greatly appreciated! An example input file is detailed below.

Thanks very much

units real
boundary p p p
newton on # use off for np>4

atom_style molecular
bond_style hybrid harmonic class2
angle_style hybrid harmonic poly4
dihedral_style hybrid multi/harmonic multi/harmonic8
improper_style hybrid harmonic
pair_style hybrid/overlay lj96/cut 20.0 dpd/tstat 353.0 353.0 20.0 5839988

read_data 3ht10_4_cg.dat

Lennard-Jones pair coefficients: epsilon (kcal/mol), sigma (Angstrom)

pair_coeff 1 1 lj96/cut 0.4 5.0
pair_coeff 1 2 lj96/cut 0.4 5.0
pair_coeff 1 3 lj96/cut 0.4 5.0
pair_coeff 2 2 lj96/cut 0.4 5.0
pair_coeff 2 3 lj96/cut 0.4 5.0
pair_coeff 3 3 lj96/cut 0.4 5.0

dissipative particle dynamics parameters: gamma (g/mol/fs)

pair_coeff 1 1 dpd/tstat 0.002
pair_coeff 1 2 dpd/tstat 0.002
pair_coeff 1 3 dpd/tstat 0.002
pair_coeff 2 2 dpd/tstat 0.002
pair_coeff 2 3 dpd/tstat 0.002
pair_coeff 3 3 dpd/tstat 0.002

neighbor 5.0 bin
neigh_modify delay 5

communicate single vel yes

fix constnve all nve

velocity all create 353.0 3598964 mom yes dist uniform

thermo_style custom step temp pe etotal press vol epair ebond eangle edihed eimp
thermo_modify flush yes
thermo 100

dump atomdump all custom 5000 3ht1_40_cg.lammpstrj id type mol x y z ix iy iz
dump atomdump1 all custom 5000 3ht1_40_cg.atom id type mol x y z
restart 1000000 3ht1_40_cg.restart

timestep 10.0
run 2000000

Using lj96/cut and dpd/tstat for the pair style (using hybrid/overlay)

I don't know, but you do realize you are not running DPD? DPD
is pair dpd. Dpd/tstat is simply a pairwise thermostat. So you
are really running lj96 with a thermostat.