Modularising Bayesian Network (Experimental)

src/experimental/ProbabilisticGraphicalModels/ProbabilisticGraphicalModels.jl

@@ -4,17 +4,15 @@ using BangBang
 using Graphs
 using Distributions
 
-include("bayesnet.jl")
+include("bayesian_network.jl")
+include("conditioning.jl")
+include("functions.jl")
 
 export BayesianNetwork,
-    add_stochastic_vertex!,
-    add_deterministic_vertex!,
-    add_edge!,
     condition,
     condition!,
     decondition,
     decondition!,
     ancestral_sampling,
     is_conditionally_independent
-
 end

src/experimental/ProbabilisticGraphicalModels/bayesian_network.jl

@@ -0,0 +1,39 @@
+"""
+    BayesianNetwork
+
+A structure representing a Bayesian Network.
+"""
+struct BayesianNetwork{V,T,F}
+    graph::SimpleDiGraph{T}
+    "names of the variables in the network"
+    names::Vector{V}
+    "mapping from variable names to ids"
+    names_to_ids::Dict{V,T}
+    "values of each variable in the network"
+    values::Dict{V,Any} # TODO: make it a NamedTuple for better performance in the future
+    "distributions of the stochastic variables"
+    distributions::Vector{Distribution}
+    "deterministic functions of the deterministic variables"
+    deterministic_functions::Vector{F}
+    "ids of the stochastic variables"
+    stochastic_ids::Vector{T}
+    "ids of the deterministic variables"
+    deterministic_ids::Vector{T}
+    is_stochastic::BitVector
+    is_observed::BitVector
+end
+
+function BayesianNetwork{V}() where {V}
+    return BayesianNetwork(
+        SimpleDiGraph{Int}(), # by default, vertex ids are integers
+        V[],
+        Dict{V,Int}(),
+        Dict{V,Any}(),
+        Distribution[],
+        Any[],
+        Int[],
+        Int[],
+        BitVector(),
+        BitVector(),
+    )
+end