RandomWalk(Diffusion) Covariance Kernel for GPR

Development
3

There is a mistake above. The exponential should be a matrix exponential:

i.e. tt.slinalg.expm(self.M*self.ls)