add (mostly broken, for now) tests

Author: cscherrer

test/runtests.jl

@@ -1,6 +1,148 @@
 using MeasureBase
 using Test
+using StatsFuns
+using Base.Iterators: take
+using Random
+using LinearAlgebra
 
-@testset "MeasureBase.jl" begin
-    # Write your tests here.
+function draw2(μ)
+    x = rand(μ)
+    y = rand(μ)
+    while x == y
+        y = rand(μ)
+    end
+    return (x,y)
+end
+
+@testset "Parameterized Measures" begin
+    @testset "Binomial" begin
+        D = Binomial{(:n, :p)}
+        par = merge((n=20,),transform(asparams(D, (n=20,)), randn(1)))
+        d = D(par)
+        (n,p) = (par.n, par.p)
+        logitp = logit(p)
+        probitp = norminvcdf(p)
+        y = rand(d)
+
+        ℓ = logdensity(Binomial(;n, p), y)
+        @test ℓ ≈ logdensity(Binomial(;n, logitp), y)
+        @test ℓ ≈ logdensity(Binomial(;n, probitp), y)
+
+        @test_broken logdensity(Binomial(n,p), CountingMeasure(ℤ[0:n]), x) ≈ binomlogpdf(n,p,x)
+    end
+
+    @testset "NegativeBinomial" begin
+        D = NegativeBinomial{(:r, :p)}
+        par = transform(asparams(D), randn(2))
+        d = D(par)
+        (r,p) = (par.r, par.p)
+        logitp = logit(p)
+        λ = p * r / (1 - p)
+        y = rand(d)
+
+        ℓ = logdensity(NegativeBinomial(;r, p), y)
+        @test ℓ ≈ logdensity(NegativeBinomial(;r, logitp), y)
+        @test ℓ ≈ logdensity(NegativeBinomial(;r, λ), y)
+
+        @test_broken logdensity(Binomial(n,p), CountingMeasure(ℤ[0:n]), x) ≈ binomlogpdf(n,p,x)
+    end
+
+    @testset "Normal" begin
+        D = Normal{(:μ,:σ)}
+        par = transform(asparams(D), randn(2))
+        d = D(par)
+        @test params(d) == par
+
+        μ = par.μ
+        σ = par.σ
+        σ² = σ^2
+        τ = 1/σ²
+        logσ = log(σ)
+        y = rand(d)
+
+        ℓ = logdensity(Normal(;μ,σ), y)
+        @test ℓ ≈ logdensity(Normal(;μ,σ²), y)
+        @test ℓ ≈ logdensity(Normal(;μ,τ), y)
+        @test ℓ ≈ logdensity(Normal(;μ,logσ), y)
+    end
+end
+
+@testset "Kernel" begin
+    κ = MeasureTheory.kernel(identity, MeasureTheory.Dirac)
+    @test rand(κ(1.1)) == 1.1
+end
+
+@testset "SpikeMixture" begin
+    @test rand(SpikeMixture(Dirac(0), 0.5)) == 0
+    @test rand(SpikeMixture(Dirac(1), 1.0)) == 1
+    w = 1/3
+    m = SpikeMixture(Normal(), w)
+    bm = basemeasure(m)
+    @test (bm.s*bm.w)*bm.m == 1.0*basemeasure(Normal())
+    @test density(m, 1.0)*(bm.s*bm.w) == w*density(Normal(),1.0)
+    @test density(m, 0)*(bm.s*(1-bm.w)) ≈ (1-w)
+end
+
+@testset "Dirac" begin
+    @test rand(Dirac(0.2)) == 0.2
+    @test logdensity(Dirac(0.3), 0.3) == 0.0
+    @test logdensity(Dirac(0.3), 0.4) == -Inf
+end
+
+@testset "For" begin
+    FORDISTS = [
+        For(1:10) do j Normal(μ=j) end
+        For(4,3) do μ,σ Normal(μ,σ) end
+        For(1:4, 1:4) do μ,σ Normal(μ,σ) end
+        For(eachrow(rand(4,2))) do x Normal(x[1], x[2]) end
+        For(rand(4), rand(4)) do μ,σ Normal(μ,σ) end
+    ]
+
+    for d in FORDISTS
+        @test logdensity(d, rand(d)) isa Float64
+    end
+end
+
+import MeasureTheory.:⋅
+function ⋅(μ::Normal, kernel) 
+    m = kernel(μ)
+    Normal(μ = m.μ.μ, σ = sqrt(m.μ.σ^2 + m.σ^2))
+end
+
+"""
+    ConstantMap(β)
+Represents a function `f = ConstantMap(β)`
+such that `f(x) == β`.
+"""
+struct ConstantMap{T}
+    x::T
+end
+(a::ConstantMap)(x) = a.x
+(a::ConstantMap)() = a.x
+
+struct AffineMap{S,T}
+    B::S
+    β::T
+end
+(a::AffineMap)(x) = a.B*x + a.β
+(a::AffineMap)(p::Normal) = Normal(μ = a.B*mean(p) + a.β, σ = sqrt(a.B*p.σ^2*a.B'))
+
+@testset "DynamicFor" begin
+    mc = Chain(Normal(μ=0.0)) do x Normal(μ=x) end
+    r = rand(mc)
+   
+    # Check that `r` is now deterministic
+    @test logdensity(mc, take(r, 100)) == logdensity(mc, take(r, 100))
+    
+    d2 = For(r) do x Normal(μ=x) end  
+
+    @test_broken let r2 = rand(d2)
+        logdensity(d2, take(r2, 100)) == logdensity(d2, take(r2, 100))
+    end
+end
+
+@testset "Likelihood" begin
+    d = Normal()
+    ℓ = Likelihood(Normal{(:μ,)}, 3.0) 
+    @test logdensity(d ⊙ ℓ, 2.0) == logdensity(d, 2.0) + logdensity(ℓ, 2.0)
 end