C Matrix Program

Question

I have a matrix program that so far takes an input file in and expresses the numbers in matrices. I also wrote a function to find its transpose but I can't figure out how to do the symmetrical.

I have to take only 3 input parameters (MainMatrix[max][max] MainTranspose[max][max] and the Size)

It should return -1 if it isn't symmetrical and 0 if it is.

My program shows everything right except for 1 that says it is not symmetrical when it SHOULD be.

I hope one of you guys can help me with this.

int Symmetry (int mainmatrix[max][max], int maintranspose[max][max], int size) {
    double thesqrtSize = sqrt((double)Size);
    double fract = thesqrtSize - floor(thesqrtSize);
    if(fract > 0.0001) 
        return -1;
    int isqrtSize = (int)thesqrtSize;
    int i, j;
    for(i=0;i<isqrtSize;i++){
        for(j=0;j<isqrtSize;j++) {
            if(mainmatrix[i][j]!=maintranspose[i][j]) {
                return -1; 
            }
        }
    }
    return 0; 
}

Answer 1

Floating point numbers are notoriously prone to rounding errors. And if the result of sqrt is just a little bit below the intended result, the call to floor will make it off by (almost) 1. Try adding 1 to size before taking the square root. This is safe and will guarantee a correct result provided size is a square number.

The suggested change:

double thesqrtSize = sqrt((double)(Size+1));

Obviously this means your method of using floor to check for squareness won't work. I suggest casting to int, squaring, and comparing to the original size to see if they're the same.

Answer 2

Symmetry function doesn't need the transpose matrix. You can find the answer without it:

int Symmetry (int mainmatrix[max][max], int size) {
    int isqrtSize= (int)sqrt((double)size);
    if(isqrtSize*isqrtSize!=size) 
        return -1;
    int i, j;
    for(i=0;i<isqrtSize;i++){
        for(j=0;j<isqrtSize;j++) {
            if(mainmatrix[i][j]!=mainmatrix[j][i]) {
                return -1; 
            }
        }
    }
    return 0; 
}

You didn't solve the multidimensional array parameter thing in the most elegant way, but lets use this. I think your program might work on this way:

• you declare a matrix with the maximum size
• You fill the upper left corner of the matrix and check for symmetricity
• symmetry method gets the entire matrix and try to figure out the dimensions and check the symmetry.

Now please see THIS CODE. I wrote a simple main function for this. As You can see, it works. So the problem should be at the method that generates the transpose matrix or with the given size in your program.

Another problem with the algorithm: You can not find dimensions from the number of the element. A matrix of 8*2 contains 16 element, so the method will see and handle it like it was a 4*4 matrix. If there are no cases like that, You can use the code above.

But: When you call the symmetry function, You compute the size as size=rows*columns, so You must know the sizes of the matrix. This means, You can simple give them to your function too. In this case, You don't have to do compute sqrt and other floating point values, You only have to check the input parameters if they are equal. With this, the function will look like this:

int Symmetry (int mainmatrix[max][max], int rows, int cols) {
    if(rows!=cols)
        return -1;
    int i, j;
    for(i=0;i<rows;i++){
        for(j=0;j<cols;j++) {
            if(mainmatrix[i][j]!=mainmatrix[j][i]) {
                return -1; 
            }
        }
    }
    return 0; 
}