We can rewrite the basic structure of the code in Julia and exploit the complete GPU-parallelizable environment.
This can be achieved like the C++ version of the code, but entirely in julia, to avoid PyCall problems.
The python code could export all quantities like it is done for the C++ version.
This is particularly relevant for parallelization via GPU and MPI hybrid, that could be exploited in this way.
We can rewrite the basic structure of the code in Julia and exploit the complete GPU-parallelizable environment.
This can be achieved like the C++ version of the code, but entirely in julia, to avoid PyCall problems.
The python code could export all quantities like it is done for the C++ version.
This is particularly relevant for parallelization via GPU and MPI hybrid, that could be exploited in this way.