I am wondering if someone can tell me if this does what I think it does:
HP = pm.Normal('HP', mu =0, sd= 1, shape = n2)
gm = pm.GaussianRandomWalk('gm', mu = HP, sd = 1, shape = (n1, n2))
Which is to make n2 independent random walks each of length n1, each with a drift instantiated from each element of HP?
The line runs, but I can’t find an explanation of how shape is handled here - tried looking in the code too, but
If not maybe I can do this with a MvGaussianRandomWalk, with identity covariance and mean set to HP?