Merge pull request #245 from ReactiveBayes/multivariate-input-to-f

Multivariate input to function
ReactiveBayes · Jul 3, 2024 · d83383a · d83383a
2 parents bd5f843 + efd655d
commit d83383a
Show file tree

Hide file tree

Showing 2 changed files with 55 additions and 2 deletions.
diff --git a/src/graph_engine.jl b/src/graph_engine.jl
@@ -1666,11 +1666,20 @@ function add_edge!(
     return add_edge!(model, factor_node_id, factor_node_propeties, variable_node_id, interface_name, 1)
 end
 
+add_edge!(
+    model::Model,
+    factor_node_id::NodeLabel,
+    factor_node_propeties::FactorNodeProperties,
+    variable_node_id::Union{ProxyLabel, VariableRef},
+    interface_name::Symbol,
+    index
+) = add_edge!(model, factor_node_id, factor_node_propeties, unroll(variable_node_id), interface_name, index)
+
 function add_edge!(
     model::Model,
     factor_node_id::NodeLabel,
     factor_node_propeties::FactorNodeProperties,
-    variable_node_id::Union{ProxyLabel, NodeLabel, VariableRef},
+    variable_node_id::Union{NodeLabel},
     interface_name::Symbol,
     index
 )

diff --git a/test/graph_construction_tests.jl b/test/graph_construction_tests.jl
@@ -1770,4 +1770,48 @@ end
     @test_throws "Comparing Factor Graph variable `x` with a value. This is not possible as the value of `x` is not known at model construction time." create_model(
         test_model(y = 1)
     )
-end
+end
+
+@testitem "Multivariate input to function" begin
+    using GraphPPL
+    import GraphPPL: create_model, getorcreate!, datalabel
+
+    include("testutils.jl")
+    function dot end
+    function relu end
+
+    @model function neuron(in, out)
+        local w
+        for i in 1:(length(in))
+            w[i] ~ Normal(0.0, 1.0)
+        end
+        bias ~ Normal(0.0, 1.0)
+        unactivated := dot(in, w) + bias
+        out := relu(unactivated)
+    end
+
+    @model function neural_network_layer(in, out, n)
+        for i in 1:n
+            out[i] ~ neuron(in = in)
+        end
+    end
+
+    @model function neural_net(in, out)
+        local softin
+        for i in 1:length(in)
+            softin[i] ~ Normal(in[i], 1.0)
+        end
+        h1 ~ neural_network_layer(in = softin, n = 10)
+        h2 ~ neural_network_layer(in = h1, n = 16)
+        out ~ neural_network_layer(in = h2, n = 2)
+    end
+
+    model = create_model(neural_net()) do model, ctx
+        in = datalabel(model, ctx, GraphPPL.NodeCreationOptions(kind = :data), :in, rand(3))
+        out = datalabel(model, ctx, GraphPPL.NodeCreationOptions(kind = :data), :out, randn(2))
+        return (in = in, out = out)
+    end
+    @test length(collect(filter(as_node(Normal), model))) == 253
+    @test length(collect(filter(as_node(dot), model))) == 28
+    @test length(collect(filter(as_variable(:in), model))) == 3
+end