Does "fix balance" work well when there are meshes in the domain

Dear LAMMPS developers and users,

One general question related to the “fix balance” command. Does this command re-partition computational domain ONLY for atoms/particles? If so, when I have an application in which I also have meshes in the domain, for example, in SPH or DEM simulations, this “fix balance” command should still work without any issue, right?

Thanks for any feedback on this question.

Dear LAMMPS developers and users,

One general question related to the "fix balance" command. Does this command
re-partition computational domain ONLY for atoms/particles? If so, when I

please read the documentation. we have put considerable effort into
explaining how the balancing commands work and how you can influence
their effectiveness.

have an application in which I also have meshes in the domain, for example,
in SPH or DEM simulations, this "fix balance" command should still work
without any issue, right?

i don't understand the question. yes, the command should always work.
however, whether it will improve load balance is not so easy to
predict. older versions of LAMMPS would only use the particle numbers
to adjust the domain sizes, more recently, you can also apply
different kinds of weights (and scale their impact, if needed). how
well this will apply to the kind of systems you describe, depends on
many factors and it is up to you to verify that your load imbalance
actually is reduced. LAMMPS outputs relevant information at the end of
each run.

fix balance and balance are commands that should not be used blindly,
but with great care and after first investigating and understanding
what are the contributing factors of your load balance/imbalances.

axel.

​Axel, thanks for your comments. If I want to quickly check class hierarchy and/or inheritance, is there an easy way to do that? As I searched, there is no doxygen in the lammps repo to build a class hierarchy map.

Thanks,

Axel, thanks for your comments. If I want to quickly check class hierarchy
and/or inheritance, is there an easy way to do that? As I searched, there is
no doxygen in the lammps repo to build a class hierarchy map.

given the rather flat class structure of LAMMPS and the number of
classes it has, such a doxygen generated diagram is not likely to be
useful. if you feel different, you can easily do it yourself. there is
not need to have classes decorated with doxygen comments to build such
a graph and have a doxygen output.

however, you'll probably find more digestible and helpful information
in the developer guide LAMMPS Molecular Dynamics Simulator
linked from the LAMMPS home page.

axel.