package prbnmcn-clustering

multi_start performs nstarts independent calls to the given function f, which must return some outcome together with its cost. multi_start then returns the result that has minimal cost. multi_start_parallel does the same by forking several processes (in the hope of getting some parallelism). Note: for the parallel version, the random state is reinitizalized using Random.self_init per process.