Thanks a lot for your reply. In your case you had random forces involved
in pressure claculation. As far as I could figure out from pair_dpd.cpp
file the drag is not random. As I understand in my problem I don't have any
random forces therefore the pressure should not jump at restart.
If I remember correctly, the discontinuities in thermodynamic quantities were
not caused by the randomness of the force, but by the fact that the forces
depended on the particle velocities. I am not sure if this might be the case
in your simulation as well, since you have rigid body constraints.
Just out of curiosity, could you check if those discontinuities arise only
after a restart command, or also when simply executing multiple "run"
commands after another?