#include "mpi.h"
/* combined mpi test program */
#include <stdio.h>
#include <string.h>
#include <math.h>

double f( double a )
{
    return (4.0 / (1.0 + a*a));
}


int main(int argc, char ** argv) {
  int rank;
  int size;
  int source;
  int dest;
  int tag=50;
#define BUFFERSIZE 128
  char message[BUFFERSIZE]; 
  MPI_Status status;
  MPI_Request request;
  int  namelen;
  char processor_name[MPI_MAX_PROCESSOR_NAME];

  MPI_Init(&argc, &argv);
  MPI_Comm_rank(MPI_COMM_WORLD, &rank);  
  MPI_Comm_size(MPI_COMM_WORLD, &size);
  MPI_Get_processor_name(processor_name,&namelen);
  printf("Process %d on %s\n",rank, processor_name);

  printf("\ntesting MPI_Bcast from node zero:\n");
  if( rank == 0 )
    sprintf(message,"Hello MPI World from processor 0");
  else
    sprintf(message,"no message received");
  MPI_Bcast(message,BUFFERSIZE, MPI_CHAR, 0, MPI_COMM_WORLD); 
  printf("The message on node %d says \"%s\"\n", rank, 
         message); 

  printf("\ntesting MPI_Bcast from node 3:\n");
  if( rank == 3 )
    sprintf(message,"Hello MPI World from processor 3");
  else
    sprintf(message,"no message received");
  MPI_Bcast(message,BUFFERSIZE, MPI_CHAR, 3, MPI_COMM_WORLD); 
  printf("The message on node %d says \"%s\"\n", rank, 
         message); 

  printf("\ntesting MPI_Send/Recv:\n");
  /* MPI_Send and MPI_Recv should be separated by an MPI_Wait on each node */
  sprintf(message, "Greetings from process %d using MPI_Send/Recv", rank);
  dest = 0;
  for (source=0; source<size;source++) {
    if (rank == source){
      MPI_Send(message, strlen(message)+1,
	       MPI_CHAR, dest, tag, MPI_COMM_WORLD);
    }

    MPI_Wait( &request, &status);

    if (rank == 0){
      MPI_Recv(message,100,MPI_CHAR,source,tag,
	       MPI_COMM_WORLD, &status);
      printf("%s\n",message);
    }
  }

  printf("\ntesting MPI_Isend/Irecv:\n");
  /* MPI_Isend and MPI_Irecv have the MPI_Wait appear after the Irecv call */
  sprintf(message, "Greetings from process %d using MPI_Isend/Irecv", rank);
  dest = 0;

  for (source=0; source<size ;source++) {

    if (rank == 0){
      MPI_Irecv(message,100,MPI_CHAR,source,tag,
		MPI_COMM_WORLD, &request);
    }

    if (rank == source){
      MPI_Isend(message, strlen(message)+1,
		MPI_CHAR, dest, tag, MPI_COMM_WORLD,&request);
    }

    MPI_Wait( &request, &status);

    if (rank == 0){
      printf("%s\n",message);
    }
  }

  printf("\ntesting MPI_Reduce:\n");
  /* calculates pi by integrating 1/(1+x^2), giving each processor
     a part of the integral */
  int done = 0, n, i=0;
  double PI25DT = 3.141592653589793238462643;
  double mypi, pi, h, sum, x;
  double startwtime, endwtime;
  n = 0;
  while (!done)
    {
      startwtime = MPI_Wtime();
      if (rank == 0)
	{
	  if (n==0) n=10000; else n=0;
	}
      MPI_Bcast(&n, 1, MPI_INT, 0, MPI_COMM_WORLD);
      if (n == 0)
	done = 1;
      else
	{
	  h   = 1.0 / (double) n;
	  sum = 0.0;
	  for (i = rank + 1; i <= n; i += size)
	    {
	      x = h * ((double)i - 0.5);
	      sum += f(x);
	    }
	  mypi = h * sum;

	  MPI_Reduce(&mypi, &pi, 1, MPI_DOUBLE, MPI_SUM, 0, MPI_COMM_WORLD);

	  if (rank == 0)
	    {
	      printf("pi is approximately %.16f, Error is %.16f\n",
		     pi, fabs(pi - PI25DT));
	      endwtime = MPI_Wtime();
	      printf("wall clock time = %f\n",
		     endwtime-startwtime);           
	    }
	}
    }

  printf("\ntesting MINLOC reduction\n");
  i=rank-37;
  i*=i;
  i+=7;
  * (int *) message=i;
  MPI_Allreduce(&message,&message,1,MPI_2INT,MPI_MINLOC,MPI_COMM_WORLD);
  i=* (int *) message;
  n=* (int *) (message+sizeof(int));
  printf ("min of (processor-37)^2+7 is %d and occurs on node %d\n",i,n);

  MPI_Finalize();
  return 0;
}