I don't think there is a way right now to disable this, as it is automatically called upon initialization. In the past, I have just commented these lines locally or resorted to stdout hacks like this one.

Maybe meep could check for environment variables or some global config on initialization, something like MEEP_VERBOSITY_LEVEL or a .meeprc?

Another idea would be to use logging and replace all the prints scattered throughout the code (the adjoint module also does this without a way to disable it). This would probably be a useful thing to do anyways and could be neatly integrated with the existing verbosity system.

So as you know, meep has a verbosity setting. Unfortunately, the mpi output is triggered before any verbosity can be registered, and the adjoint outputs simply aren't wrapped with the verbosity class yet (the adjoint solver still isn't "documented" so there are several things like verbosity that aren't yet consistent with the main meep package).

An env variable may be worth it just for the mpi output (it's certainly easy to implement and add upstream). You'd have to set this env variable in gdsfactory, as we'd probably want the default behavior to stay consistent with master.

You want to throw a quick PR together and we can see how it fits?

Regarding the adjoint solver verbosity - it's trivial to add a check for verbosity.meep to the prints, but this would mean that for verbosity.meep = 0 you would not get the Starting forward/adjoint... messages anymore, which one might still want to be able to see without the simulation "clutter". The solutions I see are to either add another property to the verbosity manager (e.g. verbosity.adjoint) or perhaps add more fine-grained verbosity levels to the python module. What do you think?