Issue using MultiProcess engine about pypesto HOT 2 CLOSED

Hi @yannikschaelte I found the issue after some digging. The issue lies with the version of libroadrunner and has nothing to do with pyPESTO or AMICI.

Thank you for the helpful explanation and my apologies for wasting your time.

Thanks,
Caroline

from pypesto.

Hi @LarkinIt , in order to parallelize, the MultiProcessEngine needs to copy the problem over to other threads, which is performed by "pickling" in Python. While most Python-only objects are usually pickleable out of the box, this gets trickier for other objects, which may then need to implement own __getstate__ and __setstate__ methods. If you use AMICI via the AmiciObjective, it should be pickleable already. For tellurium, we have currently no such wrapper. If you need to use a custom AMICI based objective not based on AmiciObjective, or another simulation framework, the AmiciObjective implementation of said methods https://github.com/ICB-DCM/pyPESTO/blob/develop/pypesto/objective/amici/amici.py#L252 may still be useful (which essentially recreate the problem from file).

from pypesto.