Different results using 2 diff version of lammps

Hi Lammps Users,

I am using Granular package.
I would like to read in initial velocities with line

read_data sphere atom-ID atom-type diameter density x y z
and go something like
sphere atom-ID atom-type diameter density x y z vx vy vz

Is it possible to read in velocity/forces and impose on the atoms as their initial state?

I guess read_restart does the similar job from previous state of a simulation, but I am trying to compare 2 different versions of lammps (21May08 and 22feb2013) and need input file to read in ( cant use same restart file for both)

Source of the question is :
I am finding different results for sedimentation of particles with 2 different versions stated. and I am trying to find why even though all the parameters are same. I read data from a file and allow them to fall under gravity, after say 0.1 seconds, I can see subtle differences in the particle positions and this changes more if i allow to go for 1 sec simulation.

My input script is :

2D particle pipe simulation

atom_style sphere
atom_modify map array
boundary f f f
newton off

communicate single vel yes
read_data IC11880finalbed.in
group FIXED type 3
group MOVE type 1
velocity FIXED set 0.0 0.0 0.0 units box

neighbor 5.0e-4 bin
neigh_modify delay 0

pair_style gran/hooke/history 200.0 NULL 270.4 NULL 0.1 0
pair_coeff * *

timestep 1.0e-6

fix 1 MOVE nve/sphere
fix 2 MOVE fdrag
fix 3 MOVE gravity 9.8 vector 0 -1 0 # spherical 90.0 -180.0

fix xwall all wall/gran 200.0 NULL 192.0 NULL 0.1 0 xplane 0.00 0.044
fix ywall all wall/gran 200.0 NULL 192.0 NULL 0.1 0 yplane -0.06 0.12
fix zwall all wall/gran 200.0 NULL 192.0 NULL 0.1 0 zplane 0.00 0.01

#thermo_style granular # not work. syntax change?
thermo_style one # granular does not work
thermo 2000
thermo_modify lost error
dump id all custom 100000 snapshot.bubblemd id type x y z vx vy vz fy
#restart 1000000 biDisperse1.2mB/restart.*.bubblemd

Suggestions and insights most welcome.

Best wishes
Prashant

Hi Lammps Users,

I am using Granular package.
I would like to read in initial velocities with line

read_data sphere atom-ID atom-type diameter density x y z
and go something like
sphere atom-ID atom-type diameter density x y z vx vy vz

Is it possible to read in velocity/forces and impose on the atoms as their
initial state?

it doesn't work like that. the read_data documentation describes the
format including the velocities sections.

I guess read_restart does the similar job from previous state of a
simulation, but I am trying to compare 2 different versions of lammps
(21May08 and 22feb2013) and need input file to read in ( cant use same
restart file for both)

Source of the question is :
I am finding different results for sedimentation of particles with 2
different versions stated. and I am trying to find why even though all the

what is a "different result". you can hardly expect the results to be
numerically equal. numerical simulations based on coupled partial
differential equations like MD are chaotic systems and thus diverge
exponentially.

parameters are same. I read data from a file and allow them to fall under
gravity, after say 0.1 seconds, I can see subtle differences in the particle
positions and this changes more if i allow to go for 1 sec simulation.

why should that not be so? see above. also, in 5 years there may be
the odd bug that was fixed and enhancement that was added. on the
contrary, it would be surprising if the results would be bitwise
identical.

axel.