fixed error in BasicPOMCP minimal example

JuliaPOMDP · Nov 13, 2018 · 435d745 · 435d745
1 parent 09160cf
commit 435d745
Show file tree

Hide file tree

Showing 2 changed files with 38 additions and 16 deletions.
diff --git a/src/pomdps.jl b/src/pomdps.jl
@@ -52,18 +52,35 @@ end
 
 function initialize_belief(up::BasicParticleFilter, d::D) where D
     # using weighted iterator here is more likely to be order n than just calling rand() repeatedly
-    # but, this implementation may change in the future
-    if @implemented(support(::D)) && @implemented(pdf(::D, ::typeof(first(support(d)))))
-    # if @implemented(weighted_iterator(::D))
-        S = typeof(first(support(d)))
-        particles = S[]
-        weights = Float64[]
-        for (s, w) in weighted_iterator(d)
-            push!(particles, s)
-            push!(weights, w)
+    # but, this implementation is problematic and may change in the future
+    try
+        if @implemented(support(::D)) &&
+            @implemented(iterate(::typeof(support(d)))) &&
+            @implemented(pdf(::D, ::typeof(first(support(d)))))
+                S = typeof(first(support(d)))
+                particles = S[]
+                weights = Float64[]
+                for (s, w) in weighted_iterator(d)
+                    push!(particles, s)
+                    push!(weights, w)
+                end
+                return resample(ImportanceResampler(up.n_init), WeightedParticleBelief(particles, weights), up.rng)
+        end
+    catch ex
+        if ex isa MethodError
+            @warn("""
+                Suppressing MethodError in initialize_belief in ParticleFilters.jl. Please file an issue here:
+                
+                https://github.com/JuliaPOMDP/ParticleFilters.jl/issues/new
+
+                The error was
+
+                $(sprint(showerror, ex))
+                  """, maxlog=1)
+        else
+            rethrow(ex)
         end
-        return resample(ImportanceResampler(up.n_init), WeightedParticleBelief(particles, weights), up.rng)
-    else
-        return ParticleCollection(collect(rand(up.rng, d) for i in 1:up.n_init))
     end
+
+    return ParticleCollection(collect(rand(up.rng, d) for i in 1:up.n_init))
 end
diff --git a/test/runtests.jl b/test/runtests.jl
@@ -5,8 +5,7 @@ using Test
 using POMDPPolicies
 using POMDPSimulators
 using Random
-import ParticleFilters: obs_weight
-import POMDPs: observation
+using Distributions
 using NBInclude
 
 struct P <: POMDP{Nothing, Nothing, Nothing} end
@@ -15,7 +14,7 @@ struct P <: POMDP{Nothing, Nothing, Nothing} end
     @test !@implemented obs_weight(::P, ::Nothing, ::Nothing, ::Nothing)
     @test !@implemented obs_weight(::P, ::Nothing, ::Nothing)
 end
-obs_weight(::P, ::Nothing, ::Nothing, ::Nothing) = 1.0
+ParticleFilters.obs_weight(::P, ::Nothing, ::Nothing, ::Nothing) = 1.0
 
 @testset "implemented" begin
     @test @implemented obs_weight(::P, ::Nothing, ::Nothing, ::Nothing)
@@ -24,12 +23,13 @@ obs_weight(::P, ::Nothing, ::Nothing, ::Nothing) = 1.0
     @test obs_weight(P(), nothing, nothing, nothing, nothing) == 1.0
 end
 
-observation(::P, ::Nothing) = nothing
+POMDPs.observation(::P, ::Nothing) = nothing
 @test @implemented obs_weight(::P, ::Nothing, ::Nothing)
 
 include("example.jl")
 include("domain_specific_resampler.jl")
 
+struct ContinuousPOMDP <: POMDP{Float64, Float64, Float64} end
 @testset "infer" begin
     p = TigerPOMDP()
     filter = SIRParticleFilter(p, 10000)
@@ -70,6 +70,11 @@ include("domain_specific_resampler.jl")
         wp2 = @inferred collect(weighted_particles(WeightedParticleBelief([1,2], [0.5, 0.5])))
         @test wp1 == wp2
     end
+
+    @testset "normal" begin
+        pf = SIRParticleFilter(ContinuousPOMDP(), 100)
+        ps = @inferred initialize_belief(pf, Normal())
+    end
 end
 
 @testset "alpha" begin