package gpr

Covariance of linear functions with one hyperparameter

The covariance is defined as:

k(x, y) = x*inv(P)*y + 1/t^2 logtheta = log(t)

where P is a diagonal matrix containing t^2 along the diagonal.