Fix graph generation

docs/literate.jl

@@ -15,4 +15,4 @@ using Literate: Literate
 
 # Convert to markdown and notebook
 const SCRIPTJL = joinpath(EXAMPLEPATH, "script.jl")
-Literate.markdown(SCRIPTJL, OUTDIR; name=EXAMPLE, documenter=false, execute=true)
+Literate.markdown(SCRIPTJL, OUTDIR; name=EXAMPLE, execute=true)

docs/make.jl

@@ -61,7 +61,7 @@ makedocs(;
             )...,
         ],
     ],
-    strict=true,
+    #strict=true,
     checkdocs=:exports,
     doctestfilters=[
         # Older versions will show "0 element Array" instead of "Type[]".

examples/gaussian-ssm/script.jl

@@ -83,12 +83,10 @@ for t in 1:Tₘ
 end
 
 # Here are the latent and obseravation timeseries
-plot(x; label="x")
-xlabel!("t")
+plot(x; label="x", xlabel="t")
 
-#  
-plot(y; label="y")
-xlabel!("x")
+# 
+plot(y; label="y", xlabel="t")
 
 # `AdvancedPS` subscribes to the `AbstractMCMC` API. To sample we just need to define a Particle Gibbs kernel
 # and a model interface. 
@@ -103,11 +101,9 @@ mean_trajectory = mean(particles; dims=2);
 #md nothing #hide
 
 # This toy model is small enough to inspect all the generated traces:
-scatter(particles; label=false, opacity=0.01, color=:black)
+scatter(particles; label=false, opacity=0.01, color=:black, xlabel="t", ylabel="state")
 plot!(x; color=:darkorange, label="Original Trajectory")
 plot!(mean_trajectory; color=:dodgerblue, label="Mean trajectory", opacity=0.9)
-xlabel!("t")
-ylabel!("State")
 
 # We used a particle gibbs kernel with the ancestor updating step which should help with the particle 
 # degeneracy problem and improve the mixing. 
@@ -116,7 +112,5 @@ update_rate = sum(abs.(diff(particles; dims=2)) .> 0; dims=2) / Nₛ
 #md nothing #hide
 
 # and compare it to the theoretical value of $1 - 1/Nₚ$. 
-plot(update_rate; label=false, ylim=[0, 1], legend=:bottomleft)
+plot(update_rate; label=false, ylim=[0, 1], legend=:bottomleft, xlabel="Iteration", ylabel="Update rate")
 hline!([1 - 1 / Nₚ]; label="N: $(Nₚ)")
-xlabel!("Iteration")
-ylabel!("Update rate")

examples/particle-gibbs/script.jl

@@ -67,12 +67,11 @@ for t in 1:Tₘ
 end
 
 # Here are the latent and observation series:
-plot(x; label="x")
-xlabel!("t")
+plot(x; label="x", xlabel="t")
+
+# 
+plot(y; label="y", xlabel="t")
 
-#  
-plot(y; label="y")
-xlabel!("x")
 
 # Each model takes an `AbstractRNG` as input and generates the logpdf of the current transition:
 function (model::NonLinearTimeSeries)(rng::Random.AbstractRNG)
@@ -125,20 +124,15 @@ mean_trajectory = mean(particles; dims=2)
 # We can now plot all the generated traces.
 # Beyond the last few timesteps all the trajectories collapse into one. Using the ancestor updating step can help 
 # with the degeneracy problem.
-plot()
-scatter(particles; label=false, opacity=0.01, color=:black)
+scatter(particles; label=false, opacity=0.01, color=:black, xlabel="t", ylabel="state")
 plot!(x; color=:darkorange, label="Original Trajectory")
 plot!(mean_trajectory; color=:dodgerblue, label="Mean trajectory", opacity=0.9)
-xlabel!("t")
-ylabel!("State")
 
 # We can also check the mixing as defined in the Gaussian State Space model example. As seen on the
 # scatter plot above, we are mostly left with a single trajectory before timestep 150. The orange 
 # bar is the optimal mixing rate for the number of particles we use.
 update_rate = sum(abs.(diff(particles; dims=2)) .> 0; dims=2) / Nₛ
 #md nothing #hide
 
-plot(update_rate; label=false, ylim=[0, 1], legend=:bottomleft)
+plot(update_rate; label=false, ylim=[0, 1], legend=:bottomleft, xlabel="Iteration", ylabel="Update rate")
 hline!([1 - 1 / Nₚ]; label="N: $(Nₚ)")
-xlabel!("Iteration")
-ylabel!("Update rate")