I have a multi-dimensional fit with sometimes 100s of free parameters. Minuit is doing the job but it takes a long time. Googleing around I see people have talked about MPI or openmp etc (for which I have next to no experience). Alfio Lazzaro and Lorenzo Moneta 2010 J. Phys.: Conf. Ser. 219 042044 for example.
Can anyone point me to an example of how to run a minuit minimization on multiple cores at once to get the minimization done faster. All my code is c++ and running compiled.
You can use multi-threads (but requires the function to be minimized to be thread safe) with OpenMP or multi-process with MPI.
FOr doing this you need to build minuit2 with the cmake option -Dminuit2_omp=On in the first case and have openMP installed in your system or -Dminuit2_mpi=On in the second case but you need to have installed MPI.
An example for running Minuit in parallel is this Minuit2 test program