rerun and initial run

Thank you so much for the suggestion.
The numbers of digits in the coordinates of the dump file affect the compute of the rerun.

dump 1 all custom 1 coh.dump id x y z vx vy vz ix iy iz
It is likely that the numbers of digits in this dump without dump_modify are not uniform and fixed.
3833 23.442 18.4482 -5.11832 3.44937 0.906005 -6.06541 0 0 0
3834 26.98 17.2321 -1.67285 -2.31784 4.78301 6.61861 0 0 0
3835 33.3827 30.734 -1.78418 -0.998782 -2.19827 -2.71873 0 0 0

dump_modify 1 format “%d %10.7f %10.7f %10.7f %10.7f %10.7f %10.7f %d %d %d”
After modifying(increasing) the number of digits in the coordinates of the dump file, the error of the rerun goes down.

Maybe the data in the dump file could not be the same as what the initial run used. Is it right?

Best regards,
Erchia

After modifying(increasing) the number of digits in the coordinates of the dump file, the error of the rerun goes down.

That’s a signature of round-off effects. So it’s not an “error”,

it’s just the way MD works. Even if you read a restart file (binary)

instead of a dump file (text), you would see the same effect

after enough timesteps, e.g. if you run on a different # of processors.

Steve