Autocorrelation in a model

Questions
8

Just to be clear:

data = pm.AR('y', 
                 Ro, 
                 tau=tauRecruit / Omega / Omega, 
                 init=pm.Normal.dist(0., tau=tauRecruit), 
                 shape=nyear)

is the LogRecruit as in the pymc2 code.

So yes, you can generate two random arrays similarly with pm.AR and multiply them.

The example above is conditioned on an observed, but pm.AR works similar when you are treating it as a latent Variable following some distribution. See the docstring: http://docs.pymc.io/api/distributions/timeseries.html#pymc3.distributions.timeseries.AR