The unwrapped coordinates of some atoms change dramatically at certain time steps under shear conditions

[Lammps_beginner]

Dear users, I encounter a strange phenomenon. I manage to the simulation of polymer melts at a fixed shear rate under periodic boundary conditions. However, I find that in the dump file of some steps, some atoms’ unwrapped coordinates have a dramatic change, as shown in the figure. The header line is “ITEM: ATOMS id type xu yu zu vx vy vz”.

My shear and dump commands are

fix 1 all nvt/sllod temp 1 1 1

fix 2 all deform 1 xy erate 1.979797979797979e-04 units box remap v

dump 6 all custom 100000 stage6.*.${simname}.xyz id type xu yu zu vx vy vz

respectively.
Could you please tell me whether I can adjust the atomic coordinates by using periodic boundary conditions or there are other mistakes? I would very appreciate it if you could offer some guidance about this question.
Best wishes!

[akohlmey]

You are missing the most important information here: which LAMMPS version do you use?

Also, please note that the fix deform documentation discusses box flips if the system becomes too “flat” and thus computation inefficient and eventually impossible to continue, so you should check if the jumps you see correspond to the box flips.

[Lammps_beginner]

Extreme thanks to your reply!
My Lammps version is “stable-29Aug2024_update2”.
I am afraid that because i want to put forward a perhaps naive question-- If the issue is due to box flips, why only few atoms jumps at the same time?
Thanks again for your early reply!