be more permissive with what gets inserted into categorical vectors

src/categorical_vector.jl

@@ -2,14 +2,14 @@ mutable struct CategoricalVector{T}
     items::Vector{T}
     cdf::Vector{Float64}
 
-    CategoricalVector{T}(item::T, weight::Float64) where T = new(T[item], Float64[weight])
+    CategoricalVector{T}(item::T, weight) where T = new(T[item], Float64[weight])
 end
 
-CategoricalVector(item::T, weight::Float64) where T = CategoricalVector{T}(item, weight)
+CategoricalVector(item::T, weight) where T = CategoricalVector{T}(item, weight)
 
 n_items(d::CategoricalVector) = length(d.items)
 
-function insert!(c::CategoricalVector{T}, item::T, weight::Float64) where T
+function insert!(c::CategoricalVector, item, weight)
     push!(c.items, item)
     push!(c.cdf, c.cdf[end]+weight)
 end

test/categorical_tree.jl

@@ -1,6 +1,3 @@
-using POMCPOW
-using Base.Test
-
 t = CategoricalTree(1, 1.0)
 @test POMCPOW.nleaves(t) == 1
 insert!(t, 2, 3.0)

test/categorical_vector.jl

@@ -1,25 +1,25 @@
-using POMCPOW
-using Base.Test
-
 t = CategoricalVector(1, 1.0)
 insert!(t, 2, 3.0)
-rand(Base.GLOBAL_RNG, t)
-
-results = Int[]
-@time for i in 1:1000
-    push!(results, rand(Base.GLOBAL_RNG, t))
-end
+rand(Random.default_rng(), t)
 
-@time for i in 3:1000
-    insert!(t,i,1.0)
-end
-
-@time for i in 1:1000
-    push!(results, rand(Base.GLOBAL_RNG, t))
-end
+# results = Int[]
+# @time for i in 1:1000
+#     push!(results, rand(Random.default_rng(), t))
+# end
+# 
+# @time for i in 3:1000
+#     insert!(t,i,1.0)
+# end
+# 
+# @time for i in 1:1000
+#     push!(results, rand(Random.default_rng(), t))
+# end
 
 #=
 using Plots
 histogram(results)
 gui()
 =#
+
+t2 = CategoricalVector(2, 1.0)
+insert!(t2, 2.0, 3) # test that types can get converted correctly

test/init_node_sr_belief_error.jl

@@ -24,7 +24,7 @@ function POMDPs.reward(m::SimplePOMDP, s::Int, a::Int, sp::Int)
     end
 end
 
-function POMDPs.initialstate_distribution(m::SimplePOMDP)
+function POMDPs.initialstate(m::SimplePOMDP)
     return SparseCat(1:7, ones(7) ./ 7)
 end
 

test/runtests.jl

@@ -7,6 +7,8 @@ using ParticleFilters
 using POMDPTools
 using D3Trees
 
+import Random
+
 @testset "all" begin
 
     @testset "POMDPTesting" begin
@@ -86,6 +88,10 @@ using D3Trees
         @test actionvalues(planner, b) isa AbstractVector
     end
 
+    @testset "categorical vector" begin
+        include("categorical_vector.jl")
+    end
+
     @testset "init_node_sr_belief_error" begin
         include("init_node_sr_belief_error.jl")
     end;