/* file: column_move.c
   compile: cc column_move.c -lmpi
   run: mpirun -np 2 ./a.out
*/

#include "mpi.h" 
main(int argc, char* argv[]){
    int myrank;
    int i,j;
    int dest=1;
    int source=0;
    int buf[3][6];
    MPI_Status status;
    MPI_Datatype newtype;

    MPI_Init(&argc, &argv); 
    MPI_Comm_rank(MPI_COMM_WORLD, &myrank); 

    if(myrank==0){  
      /* Processor0 
           rank = 0
           11 12 13 14 15 16 
           21 22 23 24 25 26 
           31 32 33 34 35 36
      */	   
      for(i=0; i<3; i++){
	for(j=0; j<6; j++){
	  buf[i][j]=10*(i+1)+(j+1);
	}
      }
      printf("rank = %d\n",myrank);
      for(i=0; i<3; i++){
	for(j=0; j<6; j++){
	  printf("%d ", buf[i][j]);
	}
	printf("\n");
      }
    }
    else{
      /* Processor1
           rank = 1
           0  0  0  0  0  0 
           0  0  0  0  0  0 
           0  0  0  0  0  0 
      */
      for(i=0; i<3; i++){
	for(j=0; j<6; j++){
	  buf[i][j]=0;
	}
      }
    }

    MPI_Type_vector(3,1,6,MPI_INTEGER,&newtype);
    MPI_Type_commit(&newtype);

    if(myrank==0){
      MPI_Send(&buf[0][5],1,newtype,dest,100,MPI_COMM_WORLD);
    }
    else{
      MPI_Recv(&buf[0][0],1,newtype,source,100,MPI_COMM_WORLD,&status);
      /* 
       rank = 1
       16  0  0  0  0  0 
       26  0  0  0  0  0 
       36  0  0  0  0  0 
      */
      printf("rank = %d\n", myrank);
      for(i=0; i<3; i++){
	for(j=0; j<6; j++){
	  printf("%2d ", buf[i][j]);
	}
	printf("\n");
      }
    }
    MPI_Finalize(); 
  }