Defining a custom multivariate distribution from univariate distributions and stack

Questions v5
3

Thanks never looked at GaussianRandomWalk before. So if one does,

pm.GaussianRandomWalk("test", 0, np.eye(2), init_dist = pm.Normal.dist(-5,5),
                            steps=10)

does it do something like what I am asking above, say taking some normal distributions and stacking them together into a distribution of support dimension 1?