We want to share our personal experience about calling MPI fortran program from Julia and
carrying out parameter optimization on it.
Most of the time, researchers have to deal with parameter optimization of their simulation model.
Their simulation model runs generally as a parallel code written in Fortran, C or C++ using MPI libraries.
In this talk, we want to show how to connect a MPI Fortran code with Julia using MPI.jl package and ccall function
and thus benefits from all advantages of our favorite language, such as the Optim.jl package
Examples will start with simple objective functions like $\frac{1}{2}|x-c|^2$ and
and pursue with more complicated examples such as the heat equation. Some experiments on
how to manage the process optimization will be given. We will also report some demonstrations
using opencoarrays instead of MPI library.
After being graduated in applied mathematics (Ph.D. in the University of Toulouse in 2007), M. Fuentes integrated the French National Institute for Research in Computer Science and Automation (INRIA) and becomes a Research Engineer. At INRIA, his mission consists in
improving software quality trough better development practices. His main interests are
computer langages for doing maths and linear algebra