Difusion Map

Goal is to create a transition matrix M^t to jump from point x to y given the one step probability to jump p(x,y). Now, p(x,y) can be calculated using some kernel function (like Gaussian kernel function).