3.18. MPI Type vector can be used to build a derived datatype from a collection of blocks of elements in an array as long as the blocks all have the same size and they're equally spaced. Its syntax is
For example, if we had an array x of 18 doubles and we wanted to build a type corresponding to the elements in positions 0, 1, 6, 7, 12, 13, we could call int MPI Type vector(3, 2, 6, MPI DOUBLE, &vect mpi t); since the type consists of 3 blocks, each of which has 2 elements, and the spacing between the starts of the blocks is 6 doubles. Write Read vector and Print vector functions that will allow process 0 to read and print, respectively, a vector with a block-cyclic distribution. But beware! Do not use MPI Scatter or MPI Gather. There is a technical issue involved in using these functions with types created with MPI Type vector. (See, for example, [23].) Just use a loop of sends on process 0 in Read vector and a loop of receives on process 0 in Print vector. The other processes should be able to complete their calls to Read vector and Print vector with a single call to MPI Recv and MPI Send. The communication on process 0 should use a derived datatype created by MPI Type vector. The calls on the other processes should just use the count argument to the communication function, since they're receiving/sending elements that they will store in contiguous array locations.