serial_mpi

Dear lammps community

  I launched two MD simulation with same input, but one case is serial (lmp_serial), and another case is in parappel (lmp_mpi).

  I found that the trajectory of the two is not exactly the same,

# case one (mpi)
Step Temp
0 1.35
1 1.3504087
2 1.3516204
3 1.3535921
4 1.3562505
5 1.359487
6 1.3631512
7 1.3670437
8 1.3709075
9 1.3744185
10 1.3771613

# case two (serial)
Step Temp
0 1.35
1 1.3503972
2 1.3515752
3 1.3534922
4 1.3560752
5 1.3592151
6 1.3627588
7 1.3665005
8 1.3701724
9 1.3734388
10 1.3758609

    So, is this kind of difference expected or am I doing something very wrong?

Thanks

Cunzhi

your input would be required to make a rigorous judgement here, numerical error is normal but that does look pretty hefty for just ten timesteps.

Hi,

my input is shown below:

variable x equal 4
variable y equal 4
variable z equal 4

variable rho equal 0.6
variable t equal 1.35
variable rc equal 2.5

units lj
atom_style atomic

lattice fcc ${rho}
region box block 0 $x 0 $y 0 $z
create_box 1 box
create_atoms 1 box
mass 1 1.0

velocity all create $t 87287

pair_style lj/cut ${rc}
pair_coeff 1 1 1.0 1.0

neighbor 0.3 bin
neigh_modify delay 0 every 1

velocity all scale $t

fix 1 all nve

thermo 1
thermo_style custom step temp #v_Jx v_Jy v_Jz

dump 1 all custom 1 dump.lammpstrj id type x y z vx vy vz

run 10

Thanks.