Remove internal call

Author: FredericWantiez

ext/AdvancedPSLibtaskExt.jl

@@ -87,7 +87,7 @@ end
 # PG requires keeping all randomness for the reference particle
 # Create new task and copy randomness
 function AdvancedPS.forkr(trace::LibtaskTrace)
-    newf = AdvancedPS.reset_model(trace.model.ctask.fargs[1])
+    newf = AdvancedPS.reset_model(trace.model.f)
     Random123.set_counter!(trace.rng, 1)
 
     ctask = Libtask.TapedTask(trace.rng, newf)
@@ -109,7 +109,8 @@ AdvancedPS.update_ref!(::LibtaskTrace) = nothing
 Observe sample `x` from distribution `dist` and yield its log-likelihood value.
 """
 function AdvancedPS.observe(dist::Distributions.Distribution, x)
-    return Libtask.produce(Distributions.loglikelihood(dist, x))
+    Libtask.produce(Distributions.loglikelihood(dist, x))
+    return nothing
 end
 
 """
@@ -149,7 +150,7 @@ function AbstractMCMC.step(
 
     replayed = AdvancedPS.replay(newtrajectory)
     return AdvancedPS.PGSample(replayed.model.f, logevidence),
-    AdvancedPS.PGState(newtrajectory)
+    AdvancedPS.PGState(replayed)
 end
 
 function AbstractMCMC.sample(

src/container.jl

@@ -206,12 +206,12 @@ function resample_propagate!(
 
             Random.seed!(p.rng, seeds[1])
 
-            children[j += 1] = p
+            children[j+=1] = p
             # fork additional children
             for k in 2:ni
                 part = fork(p, isref)
                 Random.seed!(part.rng, seeds[k])
-                children[j += 1] = part
+                children[j+=1] = part
             end
         end
     end

test/smc.jl

@@ -7,12 +7,12 @@
         sampler = AdvancedPS.SMC(15, 0.6)
         @test sampler.nparticles == 15
         @test sampler.resampler ===
-            AdvancedPS.ResampleWithESSThreshold(AdvancedPS.resample_systematic, 0.6)
+              AdvancedPS.ResampleWithESSThreshold(AdvancedPS.resample_systematic, 0.6)
 
         sampler = AdvancedPS.SMC(20, AdvancedPS.resample_multinomial, 0.6)
         @test sampler.nparticles == 20
         @test sampler.resampler ===
-            AdvancedPS.ResampleWithESSThreshold(AdvancedPS.resample_multinomial, 0.6)
+              AdvancedPS.ResampleWithESSThreshold(AdvancedPS.resample_multinomial, 0.6)
 
         sampler = AdvancedPS.SMC(25, AdvancedPS.resample_systematic)
         @test sampler.nparticles == 25
@@ -98,7 +98,7 @@
             return AdvancedPS.observe(Bernoulli(x / 2), 0)
         end
 
-        chains_smc = sample(TestModel(), AdvancedPS.SMC(100))
+        chains_smc = sample(TestModel(), AdvancedPS.SMC(100); progress=false)
 
         @test all(isone(particle.x) for particle in chains_smc.trajectories)
         @test chains_smc.logevidence ≈ -2 * log(2)
@@ -112,12 +112,12 @@
         sampler = AdvancedPS.PG(60, 0.6)
         @test sampler.nparticles == 60
         @test sampler.resampler ===
-            AdvancedPS.ResampleWithESSThreshold(AdvancedPS.resample_systematic, 0.6)
+              AdvancedPS.ResampleWithESSThreshold(AdvancedPS.resample_systematic, 0.6)
 
         sampler = AdvancedPS.PG(80, AdvancedPS.resample_multinomial, 0.6)
         @test sampler.nparticles == 80
         @test sampler.resampler ===
-            AdvancedPS.ResampleWithESSThreshold(AdvancedPS.resample_multinomial, 0.6)
+              AdvancedPS.ResampleWithESSThreshold(AdvancedPS.resample_multinomial, 0.6)
 
         sampler = AdvancedPS.PG(100, AdvancedPS.resample_systematic)
         @test sampler.nparticles == 100
@@ -152,7 +152,7 @@
             return AdvancedPS.observe(Bernoulli(x / 2), 0)
         end
 
-        chains_pg = sample(TestModel(), AdvancedPS.PG(10), 100)
+        chains_pg = sample(TestModel(), AdvancedPS.PG(10), 100; progress=false)
 
         @test all(isone(p.trajectory.x) for p in chains_pg)
         @test mean(x.logevidence for x in chains_pg) ≈ -2 * log(2) atol = 0.01
@@ -177,7 +177,7 @@
         end
 
         pg = AdvancedPS.PG(1)
-        first, second = sample(DummyModel(), pg, 2)
+        first, second = sample(DummyModel(), pg, 2; progress=false)
 
         first_model = first.trajectory
         second_model = second.trajectory