'error: invalid operands to binary *' problem

[Kevin Breslin]

I've declared a matrix array but I've had some problems trying to solve this

float dot_product(double *, double *, size_t); 


float 
dot_product(double * weights[6][16], double * X[16], size_t n)
{
        float Net[6]={0,0,0,0,0,0};
        size_t i;
 
        for (i = 1; i < 16; i++) {
             Net[1] += weights[1][i] * X[i];
             Net[2] += weights[2][i] * X[i];
             Net[3] += weights[3][i] * X[i];
             Net[4] += weights[4][i] * X[i];
             Net[5] += weights[5][i] * X[i];
             Net[6] += weights[6][i] * X[i];
        }
        return Net[6];


}

It produces the following errors

Line 42: error: conflicting types for 'dot_product'Line 38: error: previous declaration of 'dot_product' was hereIn function 'dot_product':Line 47: error: invalid operands to binary *Line 48: error: invalid operands to binary *Line 49: error: invalid operands to binary *Line 50: error: invalid operands to binary *Line 51: error: invalid operands to binary *Line 52: error: invalid operands to binary

Any help?

[ZuK]

float dot_product(double *, double *, size_t);

is not a valid forward declaration for

float 
dot_product(double * weights[6][16], double * X[16], size_t n)
{

Kurt

[kmdv]

float dot_product(double *, double *, size_t); 


float 
dot_product(double * weights[6][16], double * X[16], size_t n)
{
        float Net[6]={0,0,0,0,0,0};
        size_t i;
 
        for (i = 1; i < 16; i++) {
             Net[1] += weights[1][i] * X[i];
             Net[2] += weights[2][i] * X[i];
             Net[3] += weights[3][i] * X[i];
             Net[4] += weights[4][i] * X[i];
             Net[5] += weights[5][i] * X[i];
             Net[6] += weights[6][i] * X[i];
        }
        return Net[6];
}

First, always keep declaration and definition same (copy-pasting is legitimate here):

float 
dot_product(double * weights[6][16], double * X[16], size_t n);

float 
dot_product(double * weights[6][16], double * X[16], size_t n)
{
    /* ... */
}

You should read more about array/pointer declarations. The first parameter "weights" is a two-dimensional array of "pointers to doubles" - not just "doubles". The second parameter isn't better.

The first dimension of array parameter is meaningless, because user can always pass an array which is larger or smaller. However, you can keep it to make it a bit self-documenting.

The thing necessary to do is to remove the asterix:

float
dot_product(double weights[][16], double X[], std::size_t n);

float
dot_product(double weights[][16], double X[], std::size_t n)
{
    /* ... */
}

[iMalc]

Arrays start at zero, not one. Valid indexes in this case are 0, 1, 2, 3, 4 and 5.

You cannot return a local array, so I don't know what you plan on having this function do.

I don't think you want any of those parameters to be pointers either.