from mpi4py import MPI
import numpy

def computeParallelFactorial(start, terms):
    nums = numpy.arange(start, start + terms, 1)
    # Local reduce
    partial = numpy.multiply.reduce(nums)
    # Global reducce
    fact = comm.allreduce(partial, op=MPI.PROD)
    return fact

if __name__ == "__main__":
    from sys import argv
    # Establish Communicator
    comm = MPI.COMM_WORLD
    # Obtain rank
    rank = comm.Get_rank()
    # Get number of processes
    np = comm.Get_size()
    terms = int(argv[1])
    if terms % np != 0:
        print("terms % np != 0")
        exit()
    l_terms = terms // np
    start = l_terms*rank+1
    result = computeParallelFactorial(start, l_terms)
    if rank == 0:
        print("{:d}! computed with {} processes is {}".format(terms, np, result))