[lammps-users] Jump in pressure at restart

Hi everyone,

  I am running a simulation of a rigid cluster consisting of 32 LJ particles in
78748 LJ solvent particles. When I restart this simulation using the restart
file written at earlier time, there is a pressure jump at the first while other
properties like temperature, energy, etc restart with exact same values. Can
anyone help me as to why this could be happening.

  Here is sample output from the log files:

I ran into the same problem in the context of DPD simulations a few months
ago, see http://sourceforge.net/mailarchive/message.php?msg_id=37606266. Paul
Crozier was kind enough to look into this issue, and if I remember correctly,
we concluded that this had something to do with whether LAMMPS uses
full-timestep or half-timestep values of the velocities to calculate
thermodynamic quantities, the latter not being available at the beginning of
a new run.

Hope this helps,


I think this is due to the estimation LAMMPS is making of the
contraint force that is applied to the rigid body which contributes
to the virial. During a running simulation (including the last step
before the restart file was written), the contraint force (and its
virial contribution) are computed in the middle of the timestep.

But when a simulation is started, the constraint has to be applied
before the 1st timestep, when the velocities are different. So the
constraint forces and virial are unavoidably a bit different.

I think, however, this just affects the virial (pressure) and not the
dynamics of the system (other than round-off).

Paul, does this sound right?


Yes, I agree with Steve's conclusion on this.