@mschauer I finally got DAGs working, at least as a prototype. Here's a tiny example:
julia> m = @model begin
p ~ Uniform()
q = 1-p
y ~ Bernoulli(q)
end;
julia> dag = getdag(m(), rand(m()))
Meta graph based on a Graphs.SimpleGraphs.SimpleDiGraph{Int64}(1, [[2], Int64[]], [Int64[], [1]]) with vertex labels of type Tuple{Symbol, Any}, vertex metadata of type Nothing, edge metadata of type Nothing, graph metadata given by nothing, and default weight 1.0
julia> dag.graph
{2, 1} directed simple Int64 graph
The rand(m()) tells it what parameters to use. We can add a method to use testvalue(m()) by default.
@mschauer I finally got DAGs working, at least as a prototype. Here's a tiny example:
The
rand(m())tells it what parameters to use. We can add a method to usetestvalue(m())by default.