Merge branch 'dev' into arithmetic

Author: ablaom

README.md

@@ -122,10 +122,10 @@ julia> pdf(v, L)
 ## Measures over finite labeled sets
 
 There is, in fact, no enforcement that probabilities in a
-`UnivariateFinite` distribution sum to one, only that they be
-non-negative. Thus `UnivariateFinite` objects can be more properly
-understood as an implementation of arbitrary non-negative measures
-over finite labeled sets.
+`UnivariateFinite` distribution sum to one, only that they be belong
+to a type `T` for which `zero(T)` is defined. In particular
+`UnivariateFinite` objects implement arbitrary non-negative, signed,
+or complex measures over a finite labeled set.
 
 
 ## What does this package provide?

src/methods.jl

@@ -330,3 +330,9 @@ function Dist.fit(d::Type{<:UnivariateFinite},
 end
 
 
+# # BROADCASTING OVER SINGLE UNIVARIATE FINITE
+
+# This mirrors behaviour assigned Distributions.Distribution objects,
+# which allows `pdf.(d::UnivariateFinite, support(d))` to work. 
+
+Broadcast.broadcastable(d::UnivariateFinite) = Ref(d)

src/types.jl

@@ -161,23 +161,14 @@ same size as the array.
 
 # # TYPES - PLAIN AND ARRAY
 
-# extend Ditributions type hiearchy to account for non-euclidean
-# supports:
-abstract type Categorical{S<:Finite} <: Dist.ValueSupport end
-
-# not exported:
-const _UnivariateFinite_{S} =
-    Dist.Distribution{Dist.Univariate,Categorical{S}}
-
 # R - reference type <: Unsigned
 # V - type of class labels (eg, Char in `categorical(['a', 'b'])`)
 # P - raw probability type
 # S - scitype of samples
 
 # Note that the keys of `prob_given_ref` need not exhaust all the
 # refs of all classes but will be ordered (LittleDicts preserve order)
-DOC_CONSTRUCTOR
-struct UnivariateFinite{S,V,R,P} <: _UnivariateFinite_{S}
+struct UnivariateFinite{S,V,R,P}
     scitype::Type{S}
     decoder::CategoricalDecoder{V,R}
     prob_given_ref::LittleDict{R,P,Vector{R}, Vector{P}}

test/methods.jl

@@ -121,6 +121,11 @@ A, S, Q, F = V[1], V[2], V[3], V[4]
     @test pdf(d, "class_1") == 0.7
 end
 
+@testset "broadcasting pdf over single UnivariateFinite object" begin
+    d = UnivariateFinite(["a", "b"], [0.1, 0.9], pool=missing);
+    @test pdf.(d, ["a", "b"]) == [0.1, 0.9]
+end 
+
 @testset "constructor arguments not categorical values" begin
     @test_throws ArgumentError UnivariateFinite(Dict('f'=>0.7, 'q'=>0.2))
     @test_throws ArgumentError UnivariateFinite(Dict('f'=>0.7, 'q'=>0.2),