To perform the reduction sum, we use the comm.Reduce statement. Also, we identify with rank zero, which is the root process that will contain recvbuf, which represents the final result of the computation:

comm.Reduce(senddata,recvdata,root=0,op=MPI.SUM) 

It makes sense to run the code with a communicator group of 10 processes, as this is the size of the manipulated array.

The output appears as follows:

C:>mpiexec -n 10 python reduction.py 
  process 1 sending [ 0 2 4 6 8 10 12 14 16 18]
on task 1 after Reduce: data = [0 0 0 0 0 0 0 0 0 0]
 process 5 sending [ 0 6 12 18 24 30 36 42 48 54]
on task 5 after Reduce: data = [0 0 0 0 0 0 0 0 0 0]
 process 7 sending [ 0 8 16 24 32 40 48 56 64 72]
on task 7 after Reduce: data = [0 0 0 0 0 0 0 0 0 0]
 process 3 sending [ 0 4 8 12 16 20 24 28 32 36]
on task 3 after Reduce: data = [0 0 0 0 0 0 0 0 0 0]
 process 9 sending [ 0 10 20 30 40 50 60 70 80 90]
on task 9 after Reduce: data = [0 0 0 0 0 0 0 0 0 0]
 process 6 sending [ 0 7 14 21 28 35 42 49 56 63]
on task 6 after Reduce: data = [0 0 0 0 0 0 0 0 0 0]
 process 2 sending [ 0 3 6 9 12 15 18 21 24 27]
on task 2 after Reduce: data = [0 0 0 0 0 0 0 0 0 0]
 process 8 sending [ 0 9 18 27 36 45 54 63 72 81]
on task 8 after Reduce: data = [0 0 0 0 0 0 0 0 0 0]
 process 4 sending [ 0 5 10 15 20 25 30 35 40 45]
on task 4 after Reduce: data = [0 0 0 0 0 0 0 0 0 0]
 process 0 sending [0 1 2 3 4 5 6 7 8 9]
on task 0 after Reduce: data = [ 0 55 110 165 220 275 330 385 440 495]