Discrepancy in results using one V.S. two dump order

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps users,

Hi,

I’m getting two different data using following codes:

fix 111 all npt temp 1 1 0.1 y 0 0 1.0 z 0 0 1.0 drag 0.3
fix 22 all deform 10 x erate 2 units box remap x
dump 1 all custom 2 dump.tensile..cfg id type xs ys zs c_stress[1] c_stress[2] c_stress[3] c_stress[4] c_stress[5] c_stress[6] c_peratom
run 5
undump 1
dump 1 all custom 1 dump.tensile..cfg id type xs ys zs c_stress[1] c_stress[2] c_stress[3] c_stress[4] c_stress[5] c_stress[6] c_peratom
run 5

fix 111 all npt temp 1 1 0.1 y 0 0 1.0 z 0 0 1.0 drag 0.3
fix 22 all deform 10 x erate 2 units box remap x
dump 1 all custom 2 dump.tensile.*.cfg id type xs ys zs c_stress[1] c_stress[2] c_stress[3] c_stress[4] c_stress[5] c_stress[6] c_peratom
run 10

using the first code my simulation size in x direction becomes 7.63 and with the second one I got 7.71 at step 10. Since fix 111 and fix 22 hasn’t been unfix through the whole simulation, why do I get different answer ? I wonder if you could please comment on it.

Regards,
Ali

2

When you do run 10 versus run 5, run 5 you will be creating

neighbor lists on different timesteps. A new run will

always reneighbor before it runs. This means particles

may get reordered on each processor. Thus there

will be round-off differences between the 2 runs and

the trajectories will diverge. It’s just a statistical

difference, not a bad vs good difference.

Steve