Implicit conversion for Vector and Matrix types

[antonott]

When using Math.NET I often find myself having to convert a scalar into a vector or matrix, to match a particular function signature. The typical example would be a class that implements a probability distribution, which can be either univariate or multivariate:

var distribution = new SomeDistribution(...);

double x = 1.3;
double y = distribution.Density(Vector<double>.Build.Dense(1, x));

...

public class SomeDistribution
{
    public double Density(Vector<double> x)
    {
        ...
    }

    ...
}

It would be nice to be able to pass the double directly in this case, and have it be automatically converted to a one-element vector. Same for matrix arguments, where scalar input could be automatically converted to a 1-by-1 matrix. Other solutions, such as overloading functions or making separate classes for the univariate cases, are cumbersome and lead to a lot of boilerplate.

I'm by no means a C# expert, but should this not be possible to implement with user-defined implicit conversion operators?

[antonott]

Also, one could similarly consider automatic casting of Vector<T> into to Matrix<T>. That is, if a function expects a matrix you should be able to call it with an n-dimensional vector, which is automatically converted to an n-by-1 matrix.

While nice in principle, I don't see this getting as much use as the main suggestion, since I expect most functions with a matrix parameter will require a square matrix. But it might be usefull for the MatrixNormal class, which I suspect most people use as a vector normal distribution anyway.