lppl
v2.0.0
|
using endog_proposal_t = pp_t<record_t<DTypes<Ts...> >&, double, Ts...> |
An endogenous proposal distribution.
An endogenous proposal distribution \(q(z|z')\) uses a past or alternate representation of the posterior distribution to construct a new representation (e.g., as is typical in generic Markov Chain Monte Carlo algorithms). An endog_proposal_t returns a double that corresponds to the log probability of the proposal distribution (e.g., loglatent(r)
).
Ts | the types of distributions in the probabilistic program |