[backcolor=rgba(252, 251, 248, 0.901961)]I am wondering how you could specify a multilevel model with an AR(1) correlation structure. For example, how do I replicate following R programming(using R package nlme) using either Jags/stan?

lme(y ~ time + x, random=~time|subject, corr=corAR1())
-----------------------------------------------------------------------------------------------

Assuming my non-hierarchical model is something like:

yi,t～N(a+b∗yi,t−1+c∗ti+d∗xi,σ2)

And the hierarchical version (grouping by subject) would be:

yi,t～N(a+b∗yi,t−1+c∗ti+d∗xi,σ2)

a～N(μ,σ2a)