DPD should work with rigid bodies

[joaander]

Original report by Jens Glaser (Bitbucket: jens_glaser, GitHub: jglaser).

---

From my recollection, the DPD force compute operates on all particles
and computes friction forces. The integrator only operates on the central particles.
Therefore computed temperature values are wrong.

The underlying reason is likely the constituent particles have no velocities. If these
were computed from the angular velocity, it should work.

A user reported this and the reproducer script is attached.

[joaander]

How would this work with MPI? It would be fairly trivial to implement velocity setting in updateCompositeParticles, but comments seem to indicate that this is run after the ghost update and that ghost constituent particles don't necessarily have their parent central particle ghost available.

[joaander]

Rescheduling to v2.5. It is not clear how to implement this.

[bdice]

This could be temporarily resolved via a documentation update that emphasizes that these two features do not currently interoperate.

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[github-actions]

This issue has been automatically closed because it has not had recent activity.

[joaander]

After the changes in #1469, this should now be possible to implement.

[joaander]

Maybe not a good first issue, as this would require adding angular momentum to the ghost particle communication.

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[github-actions]

This issue has been automatically closed because it has not had recent activity.