Template metaprogramming and floating point constant coefficients

Question

I want to know if there is way/technique in the world of C++ template metaprogramming where I can substitue hard coded floating point coeffients. To make it clearer, I will try to give an example:

Consider that I want to make a method that solves linear equations:

y = a * x + b

One way is to supply a, x, b as arguments at runtime. This I want to avoid. Consider though that I know a and b then I could write something like:

double linear(double x) {
    return 2.0 * x + 3.0;
}

So I know that a = 2.0, b= 3.0. What I want to do is to extract a and b from templates, so they are there at compile time, just as being hardcoded. An example method can look like (this is what I would like to write in my code):

template <class coefs>
double linear(double x) {
    return coefs::a * x  +  coefs::b;
}

Is this possible somehow? Not sure either if my question is clear enough so let me know if I need to rephrase.

Answer 1

Since float and double types can't be template parameters, you'll can't set your coefficients via template parameters, but you can simply hard-code the values.

For instance, this won't work:

template<float a_in, float b_in>
struct coeff
{
    static const float a = a_in;
    static const float b = b_in;
};

linear<coeff<2.0, 3.0>>(4.0);

But this will:

struct coeff
{
    static const float a = 2.0;
    static const float b = 3.0;
};

linear<coeff>(4.0);