Merge pull request #364 from ReactiveBayes/hind-on-normal-usage

Restore error message for unparametrized Normal/MvNormal/Gamma

docs/src/manuals/comparison.md

@@ -49,19 +49,19 @@ Nowadays there's plenty of probabilistic programming languages and packages avai
 using RxInfer #hide
 
 @model function inner_inner(τ, y, x)
-    y ~ Normal(τ[1], τ[2] + x)
+    y ~ Normal(mean = τ[1], var = τ[2] + x)
 end
 
 @model function inner(θ, α)
-    β ~ Normal(0, 1)
-    α ~ Gamma(β, 1)
+    β ~ Normal(mean = 0.0, var = 1.0)
+    α ~ Gamma(shape = β, rate = 1.0)
     α ~ inner_inner(τ = θ, x = 3)
 end
 
 @model function outer()
     local w
     for i = 1:5
-        w[i] ~ inner(θ = Gamma(1, 1))
+        w[i] ~ inner(θ = Gamma(shape = 1.0, rate = 1.0))
     end
     y ~ inner(θ = w[2:3])
 end

docs/src/manuals/constraints-specification.md

@@ -177,19 +177,19 @@ Read more about the `@constraints` macro in the [official documentation](https:/
 
 ```@example manual_constraints
 @model function inner_inner(τ, y)
-    y ~ Normal(τ[1], τ[2])
+    y ~ Normal(mean = τ[1], var = τ[2])
 end
 
 @model function inner(θ, α)
-    β ~ Normal(0, 1)
-    α ~ Gamma(β, 1)
+    β ~ Normal(mean = 0.0, var = 1.0)
+    α ~ Gamma(shape = β, rate = 1.0)
     α ~ inner_inner(τ = θ)
 end
 
 @model function outer()
     local w
     for i = 1:5
-        w[i] ~ inner(θ = Gamma(1, 1))
+        w[i] ~ inner(θ = Gamma(shape = 1.0, rate = 1.0))
     end
     y ~ inner(θ = w[2:3])
 end

docs/src/manuals/model-specification.md

@@ -23,7 +23,7 @@ The `@model` macro returns a regular Julia function (in this example `model_name
 ```@example model-specification-model-macro
 using RxInfer #hide
 @model function my_model(observation, hyperparameter)
-    observations ~ Normal(0.0, hyperparameter)
+    observations ~ Normal(mean = 0.0, var = hyperparameter)
 end
 ```
 

src/model/graphppl.jl

@@ -239,7 +239,7 @@ end
 
 GraphPPL.factor_alias(::ReactiveMPGraphPPLBackend, ::Type{Normal}, ::GraphPPL.StaticInterfaces{(:μ, :v)}) = ExponentialFamily.NormalMeanVariance
 GraphPPL.factor_alias(::ReactiveMPGraphPPLBackend, ::Type{Normal}, ::GraphPPL.StaticInterfaces{(:μ, :τ)}) = ExponentialFamily.NormalMeanPrecision
-GraphPPL.default_parametrization(::ReactiveMPGraphPPLBackend, ::Type{Normal}) =
+GraphPPL.default_parametrization(::ReactiveMPGraphPPLBackend, ::GraphPPL.Atomic, ::Type{Normal}, rhs) =
     error("`Normal` cannot be constructed without keyword arguments. Use `Normal(mean = ..., var = ...)` or `Normal(mean = ..., precision = ...)`.")
 
 # GraphPPL.interfaces(::ReactiveMPGraphPPLBackend, ::Type{<:ExponentialFamily.NormalMeanVariance}, _) = GraphPPL.StaticInterfaces((:out, :μ, :v))
@@ -251,15 +251,15 @@ GraphPPL.interface_aliases(::ReactiveMPGraphPPLBackend, ::Type{Normal}) = GraphP
 
 GraphPPL.factor_alias(::ReactiveMPGraphPPLBackend, ::Type{MvNormal}, ::GraphPPL.StaticInterfaces{(:μ, :Σ)}) = ExponentialFamily.MvNormalMeanCovariance
 GraphPPL.factor_alias(::ReactiveMPGraphPPLBackend, ::Type{MvNormal}, ::GraphPPL.StaticInterfaces{(:μ, :Λ)}) = ExponentialFamily.MvNormalMeanPrecision
-GraphPPL.default_parametrization(::ReactiveMPGraphPPLBackend, ::Type{MvNormal}) =
+GraphPPL.default_parametrization(::ReactiveMPGraphPPLBackend, ::GraphPPL.Atomic, ::Type{MvNormal}, rhs) =
     error("`MvNormal` cannot be constructed without keyword arguments. Use `MvNormal(mean = ..., covariance = ...)` or `MvNormal(mean = ..., precision = ...)`.")
 
 GraphPPL.interface_aliases(::ReactiveMPGraphPPLBackend, ::Type{MvNormal}) =
     GraphPPL.StaticInterfaceAliases(((:mean, :μ), (:m, :μ), (:covariance, :Σ), (:cov, :Σ), (:Λ⁻¹, :Σ), (:V, :Σ), (:precision, :Λ), (:prec, :Λ), (:W, :Λ), (:Σ⁻¹, :Λ)))
 
 GraphPPL.factor_alias(::ReactiveMPGraphPPLBackend, ::Type{Gamma}, ::GraphPPL.StaticInterfaces{(:α, :θ)}) = ExponentialFamily.GammaShapeScale
 GraphPPL.factor_alias(::ReactiveMPGraphPPLBackend, ::Type{Gamma}, ::GraphPPL.StaticInterfaces{(:α, :β)}) = ExponentialFamily.GammaShapeRate
-GraphPPL.default_parametrization(::ReactiveMPGraphPPLBackend, ::Type{Gamma}) =
+GraphPPL.default_parametrization(::ReactiveMPGraphPPLBackend, ::GraphPPL.Atomic, ::Type{Gamma}, rhs) =
     error("`Gamma` cannot be constructed without keyword arguments. Use `Gamma(shape = ..., rate = ...)` or `Gamma(shape = ..., scale = ...)`.")
 
 GraphPPL.interface_aliases(::ReactiveMPGraphPPLBackend, ::Type{Gamma}) =

test/ext/ProjectionExt/inference_with_projection_tests.jl

@@ -383,7 +383,7 @@ end
 
     @model function mymodel(y, C)
         a ~ Beta(2, 1)
-        b ~ Gamma(2, 1)
+        b ~ Gamma(shape = 2.0, rate = 1.0)
         μ := foo(a, b)
         for i in eachindex(y)
             y[i] ~ Normal(mean = μ, variance = C)
@@ -448,7 +448,7 @@ end
 
     @model function mymodel(y, C)
         a ~ Beta(1, 1)
-        b ~ Gamma(1, 1)
+        b ~ Gamma(shape = 1.0, rate = 1.0)
         μ := foo(a, b)
         for i in eachindex(y)
             y[i] ~ MvNormal(mean = μ, covariance = C)

test/inference/inference_tests.jl

@@ -30,7 +30,7 @@ end
 @testitem "__infer_create_factor_graph_model" begin
     @model function simple_model_for_infer_create_model(y, a, b)
         x ~ Beta(a, b)
-        y ~ Normal(x, 1.0)
+        y ~ Normal(mean = x, var = 1.0)
     end
 
     import RxInfer: __infer_create_factor_graph_model, ProbabilisticModel, getmodel
@@ -52,7 +52,7 @@ end
     # A simple model for testing that resembles a simple kalman filter with
     # random walk state transition and unknown observational noise
     @model function test_model1(y)
-        τ ~ Gamma(1.0, 1.0)
+        τ ~ Gamma(shape = 1.0, rate = 1.0)
 
         x[1] ~ Normal(mean = 0.0, variance = 1.0)
         y[1] ~ Normal(mean = x[1], precision = τ)
@@ -364,7 +364,7 @@ end
 @testitem "Invalid data size error" begin
     @model function test_model1(y)
         n = length(y)
-        τ ~ Gamma(1.0, 1.0)
+        τ ~ Gamma(shape = 1.0, rate = 1.0)
 
         x[1] ~ Normal(mean = 0.0, variance = 1.0)
         y[1] ~ Normal(mean = x[1], precision = τ)

test/model/graphppl_tests.jl

@@ -12,7 +12,7 @@
     input = :(a = constvar())
     @test @capture(apply_pipeline(input, error_datavar_constvar_randomvar), error(_))
 
-    input = :(x ~ Normal(0, 1))
+    input = :(x ~ Normal(mean = 0.0, var = 1.0))
     @test apply_pipeline(input, error_datavar_constvar_randomvar) == input
 end
 

test/model/initialization_plugin_tests.jl

@@ -26,8 +26,8 @@ end
     import RxInfer: SpecificSubModelInit, InitSpecification, InitDescriptor, InitMarginal, InitObject, GeneralSubModelInit
 
     @model function dummymodel()
-        x ~ Normal(0, 1)
-        y ~ Normal(x, 1)
+        x ~ Normal(mean = 0.0, var = 1.0)
+        y ~ Normal(mean = x, var = 1.0)
     end
 
     @test SpecificSubModelInit(GraphPPL.FactorID(dummymodel, 1), InitSpecification()) isa SpecificSubModelInit
@@ -42,8 +42,8 @@ end
     import RxInfer: SpecificSubModelInit, InitSpecification, InitDescriptor, InitMarginal, InitObject, GeneralSubModelInit
 
     @model function dummymodel()
-        x ~ Normal(0, 1)
-        y ~ Normal(x, 1)
+        x ~ Normal(mean = 0.0, var = 1.0)
+        y ~ Normal(mean = x, var = 1.0)
     end
 
     @test GeneralSubModelInit(dummymodel, InitSpecification()) isa GeneralSubModelInit
@@ -641,7 +641,7 @@ end
         local x
         for i in 1:3
             for j in 1:3
-                x[i, j] ~ Normal(0, 1)
+                x[i, j] ~ Normal(mean = 0.0, var = 1.0)
             end
         end
     end
@@ -683,7 +683,7 @@ end
     @test default_init(some_model) === RxInfer.EmptyInit
 
     @model function model_with_init()
-        x ~ Normal(0.0, 1.0)
+        x ~ Normal(mean = 0.0, var = 1.0)
     end
 
     default_init(::typeof(model_with_init)) = @initialization begin

test/models/aliases/aliases_gamma_tests.jl

@@ -37,3 +37,13 @@
     @test first(results.free_energy[end]) ≈ 4.385584096993327
     @test all(<=(1e-14), diff(results.free_energy)) # it oscilates a bit at the end, but all should be less or equal to zero
 end
+
+@testitem "`Gamma` by itself cannot be used as a node" begin
+    @model function gamma_by_itself(d)
+        x ~ Gamma(1.0, 1.0)
+        d ~ Gamma(x, 1.0)
+    end
+    @test_throws "`Gamma` cannot be constructed without keyword arguments. Use `Gamma(shape = ..., rate = ...)` or `Gamma(shape = ..., scale = ...)`." infer(
+        model = gamma_by_itself(), data = (d = 1.0,), iterations = 1, free_energy = false
+    )
+end

test/models/aliases/aliases_normal_tests.jl

@@ -39,3 +39,23 @@
     @test last(result.free_energy) ≈ 2.319611135721246
     @test all(iszero, diff(result.free_energy))
 end
+
+@testitem "`Normal` by itself cannot be used as a node" begin
+    @model function normal_by_itself(d)
+        x ~ Normal(0.0, 1.0)
+        d ~ Normal(x, 1.0)
+    end
+    @test_throws "`Normal` cannot be constructed without keyword arguments. Use `Normal(mean = ..., var = ...)` or `Normal(mean = ..., precision = ...)`." infer(
+        model = normal_by_itself(), data = (d = 1.0,), iterations = 1, free_energy = false
+    )
+end
+
+@testitem "`MvNormal` by itself cannot be used as a node" begin
+    @model function mvnormal_by_itself(d)
+        x ~ MvNormal(zeros(2), diageye(2))
+        d ~ MvNormal(x, diageye(2))
+    end
+    @test_throws "`MvNormal` cannot be constructed without keyword arguments. Use `MvNormal(mean = ..., covariance = ...)` or `MvNormal(mean = ..., precision = ...)`." infer(
+        model = mvnormal_by_itself(), data = (d = 1.0,), iterations = 1, free_energy = false
+    )
+end