Construct a proposal distribution over expressions.
A proposal distribution is a probability distribution conditional on an value, in our case an expression. This can be understood as the transition kernel in MCMC.
You need to define both (i) the sampler, and (ii) the procedure to compute the log conditional density.
In Julia terms this means implementing the functions:
"Generate a random candidate state e to `g(e_2 ∣ e_1 )`"
function propose(rng, e)
end
"Log density of `x1` given `x2`"
function proposallogpdf(e1, e2)
end
Construct a proposal distribution over expressions. A proposal distribution is a probability distribution conditional on an value, in our case an expression. This can be understood as the transition kernel in MCMC.
You need to define both (i) the sampler, and (ii) the procedure to compute the log conditional density. In Julia terms this means implementing the functions: