OpenMP only static scheduling with intelligent reordering of vectors by dynamic test runs TODO
mass proportional damping should be distributed on discretization and is not at the correct place (dependence on M) TODO