浮点乘法的C输出意外

Question

I am performing a multiplication and addition operation in C and am getting the wrong result.

I am performing the following operation:

#include <stdio.h>
#include <strlib.h>
#include <math.h>

int main() 
{
    float** fpimag;
    float** fpreal;

    // Some code to assign fpimag and fpreal
    // fpimag and fpreal are two dimensional arrays that are
    // passed by reference to some functions to assign data to them
    allocatememory(&fpimag,&fpreal);
    storedata(&fpimag,&fpreal);


    autofunction(&fpimag,&fpreal);
} // main()

void autofunction(float*** fpimag,float***fpreal) {
    float expreal;
    float expimag;

    expreal = cos((*afph_correct)[row][column]);
    expimag = sin((*afph_correct)[row][column]);

    (*fpimag)[row][column] = (*fpimag)[row][column] * expreal + (*fpreal)[row][column] * expimag;

 printf("Operation looks like this\n");
 printf("%f*%f + %f*%f\n",(*fpimag)[row][column],expreal,(*fpreal)[row][column],expimag);
 printf("The value is %f\n",(*fpimag)[row][column]);
} // autofunction

This is the output I get:

Operation looks like this
-0.003095*-0.431162 + 0.000027*-0.902275
The value is 0.003865

However the correct answer should be:

0.0013101

Thank you!

Answer 1

Try long double, more precise than float: http://en.wikipedia.org/wiki/Long_double

Please note that you should change your printf (%Lf for long double)

Answer 2

I'm surprised it doesn't crash. In fact I'm surprised it even compiles. I would be gobsmacked if it compiles without warnings.

• No memory is allocated for fpreal and fpimag.
• Values in fpreal are used before being initialised.
• Uses two dimensional array access but doesn't give any dimensions.
• Prints out pointers fpreal and fpimag as floats.
• Tries to print out the values as they were before the operation but value in fpimag has been overwritten.

(The first two may be covered by code not shown - the other three look to be definite problems)

Answer 3

Look at the following two lines:

(*fpimag)[row][column] = (*fpimag)[row][column] * expreal +
                         (*fpreal)[row][column] * expimag;
printf("%f*%f + %f*%f\n",fpimag,expreal,fpreal,expimag);

Can't you see that fpimag is not the same as (*fpimag)[row][column]? (and the same with fpreal) You are writing pointer to float pointer values, not the values the pointers are pointing to. Try this instead:

float imag = (*fpimag)[row][column];
float real = (*fpreal)[row][column];
(*fpimag)[row][column] = imag * expreal + real * expimag;
printf("%f*%f + %f*%f\n",imag,expreal,real,expimag);

Answer 4

Hey I found the problem it was pretty obvious:

I was overwriting the value of fpreal and then trying to use it in another calculation.

Thus I did:

(*fpreal)[row][column] = (*fpreal)[row][column]*expreal - (*fpimag)[row][column]*expimag;
(*fpimag)[row][column] = (*fpimag)[row][column]*expreal + (*fpreal)[row][column]*expimag;

I have since updated the code so that the values of fpreal and fpimag are stored in some temporary variables and use those for the calculations, as such:

oldreal = (*fpreal)[row][column];
oldimag = (*fpimag)[row][column];

(*fpreal)[row][column] = oldreal*expreal - oldimag*expimag;
(*fpimag)[row][column] = oldimag*expreal + oldreal*expimag;

Thank you everybody for your assistance though!