Hallo
I'm trying to parallelize my code. I found that it can be easily done invoking
mpirun -n 4 ./program
The program uses scotch to partition the mesh and the following assembly and solving phases, as I could understand.
I need to have access to the map of the degrees of freedom, or matrix rows, associated to each processor to perform some operations directly on the matrix. Can someone tell me how to access this information?
Alternatively, can someone tell me if there is a way to run a piece of code in serial, to perform some operations difficult to be parallelized, and the rest of the code in parallel?
Thanks for any suggestion.
