Tests green.

REQUIRE

@@ -1,2 +1,2 @@
-julia 0.6
+julia 0.7
 SpecialFunctions

src/dependency_graph.jl

@@ -115,15 +115,19 @@ Optional keyword arguments:
 This function can be used to generate message passing schedules
 if `graph` is a dependency graph.
 """
+function find_vertex_indexes(vertex::V, graph::DependencyGraph{V}) where V
+    return something(findfirst(isequal(vertex), graph.vertices), 0)
+end
+
 function children(  vertices::Vector{V},
                     graph::DependencyGraph{V};
                     allow_cycles::Bool=false,
                     breaker_sites::Set{V}=Set{V}(),
                     restrict_to::Set{V}=Set{V}()) where V
 
     # Find vertex indexes of breaker_sites
-    breaker_vertices = Set{Int}(map((v) -> something(findfirst(isequal(v), graph.vertices), 0), breaker_sites))
-    restrict_to_vertices = Set{Int}(map((v) -> something(findfirst(isequal(v), graph.vertices), 0), restrict_to))
+    breaker_vertices = Set{Int}((find_vertex_indexes(v, graph) for v=breaker_sites))
+    restrict_to_vertices = Set{Int}((find_vertex_indexes(v, graph) for v=restrict_to))
 
     visited = Int[] # Hold topologically sorted list of indices of child vertices
 

src/factor_graph.jl

@@ -113,7 +113,7 @@ end
 
 edges(graph::FactorGraph=currentGraph()) = Set{Edge}(graph.edges)
 edges(node::FactorNode) = Set{Edge}([intf.edge for intf in node.interfaces])
-edges(nodeset::Set{FactorNode}) = union(map(edges, nodeset)...)
+edges(nodeset::Set{FactorNode}) = union(Set((edges(node) for node=nodeset))...)
 
 """
 Description:

src/factor_nodes/gaussian.jl

@@ -67,7 +67,7 @@ end
 function prod!(
     x::ProbabilityDistribution{Multivariate, F1},
     y::ProbabilityDistribution{Multivariate, F2},
-    z::ProbabilityDistribution{Multivariate, GaussianWeightedMeanPrecision}=ProbabilityDistribution(Multivariate, GaussianWeightedMeanPrecision, xi=[NaN], w=[NaN].')) where {F1<:Gaussian, F2<:Gaussian}
+    z::ProbabilityDistribution{Multivariate, GaussianWeightedMeanPrecision}=ProbabilityDistribution(Multivariate, GaussianWeightedMeanPrecision, xi=[NaN], w=transpose([NaN]))) where {F1<:Gaussian, F2<:Gaussian}
 
     z.params[:xi] = unsafeWeightedMean(x) + unsafeWeightedMean(y)
     z.params[:w] = unsafePrecision(x) + unsafePrecision(y)

src/factor_nodes/gaussian_mean_precision.jl

@@ -40,7 +40,7 @@ format(dist::ProbabilityDistribution{V, GaussianMeanPrecision}) where V<:Variate
 
 ProbabilityDistribution(::Type{Univariate}, ::Type{GaussianMeanPrecision}; m=0.0, w=1.0) = ProbabilityDistribution{Univariate, GaussianMeanPrecision}(Dict(:m=>m, :w=>w))
 ProbabilityDistribution(::Type{GaussianMeanPrecision}; m::Number=0.0, w::Number=1.0) = ProbabilityDistribution{Univariate, GaussianMeanPrecision}(Dict(:m=>m, :w=>w))
-ProbabilityDistribution(::Type{Multivariate}, ::Type{GaussianMeanPrecision}; m=[0.0], w=[1.0].') = ProbabilityDistribution{Multivariate, GaussianMeanPrecision}(Dict(:m=>m, :w=>w))
+ProbabilityDistribution(::Type{Multivariate}, ::Type{GaussianMeanPrecision}; m=[0.0], w=transpose([1.0])) = ProbabilityDistribution{Multivariate, GaussianMeanPrecision}(Dict(:m=>m, :w=>w))
 
 dims(dist::ProbabilityDistribution{V, GaussianMeanPrecision}) where V<:VariateType = length(dist.params[:m])
 

src/factor_nodes/gaussian_weighted_mean_precision.jl

@@ -8,7 +8,7 @@ format(dist::ProbabilityDistribution{V, GaussianWeightedMeanPrecision}) where V<
 
 ProbabilityDistribution(::Type{Univariate}, ::Type{GaussianWeightedMeanPrecision}; xi=0.0, w=1.0) = ProbabilityDistribution{Univariate, GaussianWeightedMeanPrecision}(Dict(:xi=>xi, :w=>w))
 ProbabilityDistribution(::Type{GaussianWeightedMeanPrecision}; xi::Number=0.0, w::Number=1.0) = ProbabilityDistribution{Univariate, GaussianWeightedMeanPrecision}(Dict(:xi=>xi, :w=>w))
-ProbabilityDistribution(::Type{Multivariate}, ::Type{GaussianWeightedMeanPrecision}; xi=[0.0], w=[1.0].') = ProbabilityDistribution{Multivariate, GaussianWeightedMeanPrecision}(Dict(:xi=>xi, :w=>w))
+ProbabilityDistribution(::Type{Multivariate}, ::Type{GaussianWeightedMeanPrecision}; xi=[0.0], w=transpose([1.0])) = ProbabilityDistribution{Multivariate, GaussianWeightedMeanPrecision}(Dict(:xi=>xi, :w=>w))
 
 dims(dist::ProbabilityDistribution{V, GaussianWeightedMeanPrecision}) where V<:VariateType = length(dist.params[:xi])
 

src/probability_distribution.jl

@@ -56,7 +56,7 @@ dims(dist::ProbabilityDistribution{MatrixVariate, PointMass}) = size(dist.params
 ProbabilityDistribution(::Type{Univariate}, ::Type{PointMass}; m::Number=1.0) = ProbabilityDistribution{Univariate, PointMass}(Dict(:m=>m))
 ProbabilityDistribution(::Type{PointMass}; m::Number=1.0) = ProbabilityDistribution{Univariate, PointMass}(Dict(:m=>m))
 ProbabilityDistribution(::Type{Multivariate}, ::Type{PointMass}; m::Vector=[1.0]) = ProbabilityDistribution{Multivariate, PointMass}(Dict(:m=>m))
-ProbabilityDistribution(::Type{MatrixVariate}, ::Type{PointMass}; m::AbstractMatrix=[1.0].') = ProbabilityDistribution{MatrixVariate, PointMass}(Dict(:m=>m))
+ProbabilityDistribution(::Type{MatrixVariate}, ::Type{PointMass}; m::AbstractMatrix=transpose([1.0])) = ProbabilityDistribution{MatrixVariate, PointMass}(Dict(:m=>m))
 
 unsafeMean(dist::ProbabilityDistribution{T, PointMass}) where T<:VariateType = deepcopy(dist.params[:m])
 

src/variable.jl

@@ -44,6 +44,6 @@ end
 Collect all edges corresponding with variable(s)
 """
 edges(variable::Variable) = Set{Edge}(variable.edges)
-edges(variables::Set{Variable}) = union(map(edges, variables)...)
+edges(variables::Set{Variable}) = union(Set((edges(v) for v=variables))...)
 
 Base.isless(v1::Variable, v2::Variable) = isless("$(v1.id)", "$(v2.id)")

test/factor_nodes/test_wishart.jl

@@ -16,15 +16,15 @@ end
 
 @testset "isProper" begin
     @test isProper(ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0), nu=1.0)) == true
-    @test isProper(ProbabilityDistribution(MatrixVariate, Wishart, v=[-1.0].', nu=2.0)) == false
+    @test isProper(ProbabilityDistribution(MatrixVariate, Wishart, v=transpose([-1.0]), nu=2.0)) == false
     @test isProper(ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0), nu=0.0)) == false
 end
 
 @testset "prod!" begin
-    @test ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0), nu=2.0) * ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0),nu=2.0) == ProbabilityDistribution(MatrixVariate, Wishart, v=[0.4999999999999999].',nu=2.0)
+    @test ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0), nu=2.0) * ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0),nu=2.0) == ProbabilityDistribution(MatrixVariate, Wishart, v=transpose([0.4999999999999999]),nu=2.0)
     @test ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0), nu=2.0) * ProbabilityDistribution(MatrixVariate, PointMass, m=mat(1.0)) == ProbabilityDistribution(MatrixVariate, PointMass, m=mat(1.0))
     @test ProbabilityDistribution(MatrixVariate, PointMass, m=mat(1.0)) * ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0), nu=2.0) == ProbabilityDistribution(MatrixVariate, PointMass, m=mat(1.0))
-    @test_throws Exception ProbabilityDistribution(MatrixVariate, PointMass, m=[-1.0].') * ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0), nu=2.0)
+    @test_throws Exception ProbabilityDistribution(MatrixVariate, PointMass, m=transpose([-1.0])) * ProbabilityDistribution(MatrixVariate, Wishart, v=mat(1.0), nu=2.0)
 end
 
 @testset "unsafe mean and variance" begin
@@ -45,7 +45,7 @@ end
     @test outboundType(SPWishartOutVPP) == Message{Wishart}
     @test isApplicable(SPWishartOutVPP, [Nothing, Message{PointMass}, Message{PointMass}]) 
 
-    @test ruleSPWishartOutVPP(nothing, Message(MatrixVariate, PointMass, m=[1.0].'), Message(Univariate, PointMass, m=2.0)) == Message(MatrixVariate, Wishart, v=[1.0].', nu=2.0)
+    @test ruleSPWishartOutVPP(nothing, Message(MatrixVariate, PointMass, m=transpose([1.0])), Message(Univariate, PointMass, m=2.0)) == Message(MatrixVariate, Wishart, v=transpose([1.0]), nu=2.0)
 end
 
 @testset "VBWishartOut" begin
@@ -54,7 +54,7 @@ end
     @test isApplicable(VBWishartOut, [Nothing, ProbabilityDistribution, ProbabilityDistribution]) 
     @test !isApplicable(VBWishartOut, [ProbabilityDistribution, ProbabilityDistribution, Nothing]) 
 
-    @test ruleVBWishartOut(nothing, ProbabilityDistribution(MatrixVariate, PointMass, m=[1.5].'), ProbabilityDistribution(Univariate, PointMass, m=3.0)) == Message(MatrixVariate, Wishart, v=[1.5].', nu=3.0)
+    @test ruleVBWishartOut(nothing, ProbabilityDistribution(MatrixVariate, PointMass, m=transpose([1.5])), ProbabilityDistribution(Univariate, PointMass, m=3.0)) == Message(MatrixVariate, Wishart, v=transpose([1.5]), nu=3.0)
 end
 
 @testset "averageEnergy and differentialEntropy" begin

test/test_helpers.jl

@@ -45,7 +45,7 @@ import LinearAlgebra: Diagonal, isposdef, I, Hermitian
     @testset "diageye" begin
         # should be shorthand for Diagonal(eye(M))
         M = diageye(3)
-        @test typeof(M) == Diagonal{Float64}
+        @test typeof(M) == Diagonal{Float64, Array{Float64,1}}
         @test M == Diagonal(ones(3))
     end
 

test/test_probability_distribution.jl

@@ -54,22 +54,22 @@ end
     @test isProper(point_mass)
     @test mean(point_mass) == [0.0]
     @test var(point_mass) == [0.0]
-    @test cov(point_mass) == [0.0].'
+    @test cov(point_mass) == transpose([0.0])
 end
 
 @testset "MatrixVariate" begin
-    point_mass = ProbabilityDistribution(MatrixVariate, PointMass, m=[0.0].')
+    point_mass = ProbabilityDistribution(MatrixVariate, PointMass, m=transpose([0.0]))
     @test isa(point_mass, ProbabilityDistribution{MatrixVariate, PointMass})
-    @test point_mass.params == Dict(:m=>[0.0].')
+    @test point_mass.params == Dict(:m=>transpose([0.0]))
     @test isProper(point_mass)
-    @test mean(point_mass) == [0.0].'
+    @test mean(point_mass) == transpose([0.0])
 end
 
 @testset "PointMass ProbabilityDistribution and Message construction" begin
     @test ProbabilityDistribution(Univariate, PointMass, m=0.2) == ProbabilityDistribution{Univariate, PointMass}(Dict(:m=>0.2))
     @test_throws Exception ProbabilityDistribution(Multivariate, PointMass, m=0.2)
     @test ProbabilityDistribution(Multivariate, PointMass, m=[0.2]) == ProbabilityDistribution{Multivariate, PointMass}(Dict(:m=>[0.2]))
-    @test ProbabilityDistribution(MatrixVariate, PointMass, m=[0.2].') == ProbabilityDistribution{MatrixVariate, PointMass}(Dict(:m=>[0.2].'))
+    @test ProbabilityDistribution(MatrixVariate, PointMass, m=transpose([0.2])) == ProbabilityDistribution{MatrixVariate, PointMass}(Dict(:m=>transpose([0.2])))
     @test ProbabilityDistribution(PointMass, m=0.2) == ProbabilityDistribution{Univariate, PointMass}(Dict(:m=>0.2))
     @test ProbabilityDistribution(Univariate, PointMass) == ProbabilityDistribution{Univariate, PointMass}(Dict(:m=>1.0))
     @test ProbabilityDistribution(PointMass) == ProbabilityDistribution{Univariate, PointMass}(Dict(:m=>1.0))