Simulating Rigid Bodies with charged constituents

[amruthesht]

Hello All,

I'm trying to run rigid body simulations with charged constituent particles. I need help with the following two points.

1. I noticed that the charges key has been removed from the rigid constituent dictionary and moved to create_bodies(). Does this mean integrator.rigid = rigid wouldn't automatically assign charges to the constituents? Is there a way to continue simulating with charged constituents?
2. I understand that a GSD file doesn't store information on the constituent particles of a rigid body. However, is there a way to output the configurations actively in a format that also takes care of the constituent particle positions (so one can say visualize the simulation as a movie)?

Thank you!
Amruthesh

[joaander]

Unlike the other keys in body (type, position, and orientation): constrain.Rigid has never automatically set charges on constituents on every time step. Moving the dictionary to create_bodies reflects the actual behavior. When you call create_bodies, it will set the charges on the constituent particles that once.

GSD does store information on the constituent particles (including the charge). When you load the system state from a GSD file containing constituent particles, there is no need to call create_bodies - it is a convenience method to place particles that you could also place by other means.

To restate: If you are migrating from a previous version of HOOMD-blue, then there is no change in the behavior of constrain.Rigid. The only difference is that previously you set charges in body (which were used only once by create_bodies) and now you pass the charges to create_bodies. If you previously set constituent particles by some other means, the only change is that you no longer need to set charges in body that will go unused.

[amruthesht]

Thank you for your response.

Just to clarify, once the charges are placed using create_bodies, they will not be overwritten by the MD integrator when it is called (through integrator.rigid = rigid)(as it does for positions, orientations and typeids; explained as an example here), right?

As for outputting rigid bodies in GSD -- Is any info about the rigid body stored in the GSD, i.e., when a simulation state is initialized from a GSD, is there any info on the central particle and its corresponding constituents, or are they all treated as independent entities?

Thank you again,
Amruthesh

[joaander]

Just to clarify, once the charges are placed using create_bodies, they will not be overwritten by the MD integrator when it is called (through integrator.rigid = rigid)(as it does for positions, orientations and typeids; explained as an example here), right?

Yes, the reference docs are very clear on this point as well.

As for outputting rigid bodies in GSD -- Is any info about the rigid body stored in the GSD, i.e., when a simulation state is initialized from a GSD, is there any info on the central particle and its corresponding constituents, or are they all treated as independent entities?

GSD stores both central and constituent particles and associates them with the body tag. GSD does not store the parameters you set in rigid.body.

[amruthesht]

Thank you very much for the clarification!