/* Simple program to compute Covariance on (x,y) data */
/* Reads from two separate single-column data files   */
/* each with N rows of data (must be the same)        */
/*                                                    */
/* Written by Carey Williamson, November 2017         */
/*                                                    */
/* Usage: cc -o covariance2 covariance2.c             */
/*        ./covariance2 file1 file2                   */

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

int main(int argc, char *argv[])
  {
    float xval, yval;
    float sumx, sumy, sumxy;
    float meanx, meany, meanxy;
    int i, n;
    FILE *file1, *file2;

    if((file1 = fopen(argv[1],"r")) == NULL)
      {
	printf("Unable to open %s for input.\n", argv[1]);
	exit(1);
      }
  
    if((file2 = fopen(argv[2],"r")) == NULL)
      {
	printf("Unable to open %s for input.\n", argv[2]);
	exit(1);
      }
  
    /* Read in data values and sum them up */
    sumx = 0;
    sumy = 0;
    sumxy = 0;
    n = 0;
    while( fscanf(file1, "%f\n", &xval) == 1 )
      {
	if( fscanf(file2, "%f\n", &yval) != 1 )
	  {
	    printf("Problem with second input file!\n", argv[2]);
	    exit(1);
	  }
	sumx += xval;
	sumy += yval;
	sumxy += xval*yval;
	n++;
      }
    fclose(file1);
    fclose(file2);

    /* Compute the means of each */
    meanx = sumx/n;
    meany = sumy/n;
    meanxy = sumxy/n;

    /* Report the results */
    printf("Number of values: %d\n", n);
    printf("E[X] = %8.6f, E[Y] = %8.6f, E[XY] = %8.6f\n",
	   meanx, meany, meanxy);
    printf("Cov[XY] = E[XY] - E[X]E[Y] = %8.6f\n",
	   meanxy - meanx * meany);
  }