写一个MPI矩阵向量乘法的代码

以下是使用MPI实现矩阵向量乘法的示例代码：

#include <stdio.h>
#include <stdlib.h>
#include <mpi.h>

#define ROWS 4
#define COLS 4

int main(int argc, char *argv[]) {
    int rank, size;
    int i, j;
    int matrix[ROWS][COLS] = {{1, 2, 3, 4},
                              {5, 6, 7, 8},
                              {9, 10, 11, 12},
                              {13, 14, 15, 16}};
    int vector[COLS] = {1, 2, 3, 4};
    int result[ROWS] = {0};

    MPI_Init(&argc, &argv);
    MPI_Comm_rank(MPI_COMM_WORLD, &rank);
    MPI_Comm_size(MPI_COMM_WORLD, &size);

    if (size != COLS) {
        printf("Number of processes must match number of columns in matrix.\n");
        MPI_Finalize();
        return 1;
    }

    int *submatrix = (int *) malloc(ROWS * sizeof(int));
    MPI_Scatter(matrix, ROWS, MPI_INT, submatrix, ROWS, MPI_INT, 0, MPI_COMM_WORLD);

    int subresult = 0;
    for (i = 0; i < ROWS; i++) {
        subresult += submatrix[i] * vector[rank];
    }

    MPI_Reduce(&subresult, &result[0], 1, MPI_INT, MPI_SUM, 0, MPI_COMM_WORLD);

    if (rank == 0) {
        printf("Result: ");
        for (i = 0; i < ROWS; i++) {
            printf("%d ", result[i]);
        }
        printf("\n");
    }

    MPI_Finalize();
    return 0;
}

这个程序将矩阵和向量硬编码为整数数组，然后将矩阵的每一行分发给MPI进程。每个进程计算其分配的行和向量的乘积，并将结果通过MPI_Reduce函数发送到进程0，然后进程0打印结果。