Replace fill() with FillArrays.Fill() in product_distribution calls

Project.toml

@@ -9,6 +9,7 @@ Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 DistributionsAD = "ced4e74d-a319-5a8a-b0ac-84af2272839c"
 DynamicPPL = "366bfd00-2699-11ea-058f-f148b4cae6d8"
 Enzyme = "7da242da-08ed-463a-9acd-ee780be4f1d9"
+FillArrays = "1a297f60-69ca-5386-bcde-b61e274b549b"
 FiniteDifferences = "26cc04aa-876d-5657-8c51-4c34ba976000"
 ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
 Functors = "d9f16b24-f501-4c13-a1f2-28368ffc5196"

models/demo_assume_matrix_observe_matrix_index.jl

@@ -1,3 +1,4 @@
+using FillArrays
 using LinearAlgebra: Diagonal
 
 @model function demo_assume_matrix_observe_matrix_index(
@@ -6,7 +7,7 @@ using LinearAlgebra: Diagonal
 ) where {TV}
     n = length(x)
     d = n ÷ 2
-    s ~ reshape(product_distribution(fill(InverseGamma(2, 3), n)), d, 2)
+    s ~ reshape(product_distribution(Fill(InverseGamma(2, 3), n)), d, 2)
     s_vec = vec(s)
     m ~ MvNormal(zeros(n), Diagonal(s_vec))
     x[:, 1] ~ MvNormal(m, Diagonal(s_vec))

models/dppl_gauss_unknown.jl

@@ -1,3 +1,5 @@
+using FillArrays
+
 n = 10_000
 s = abs(rand()) + 0.5
 y = randn() .+ s * randn(n)
@@ -6,7 +8,7 @@ y = randn() .+ s * randn(n)
     N = length(y)
     m ~ Normal(0, 1)
     s ~ truncated(Cauchy(0, 5); lower = 0)
-    y ~ product_distribution(fill(Normal(m, s), N))
+    y ~ product_distribution(Fill(Normal(m, s), N))
 end
 
 model = dppl_gauss_unknown(y)

models/dppl_hier_poisson.jl

@@ -1,3 +1,4 @@
+using FillArrays
 using Turing: LogPoisson
 
 nd, ns = 5, 10
@@ -16,7 +17,7 @@ idx = repeat(collect(1:ns), inner = nd)
     a0 ~ Normal(0, 10)
     a1 ~ Normal(0, 1)
     a0_sig ~ truncated(Cauchy(0, 1); lower = 0)
-    a0s ~ product_distribution(fill(Normal(0, a0_sig), ns))
+    a0s ~ product_distribution(Fill(Normal(0, a0_sig), ns))
     alpha = a0 .+ a0s[idx] .+ a1 * x
     y ~ product_distribution(LogPoisson.(alpha))
 end

models/dppl_high_dim_gauss.jl

@@ -1,5 +1,7 @@
+using FillArrays
+
 @model function dppl_high_dim_gauss(D)
-    m ~ product_distribution(fill(Normal(0, 1), D))
+    m ~ product_distribution(Fill(Normal(0, 1), D))
 end
 
 model = dppl_high_dim_gauss(10_000)

models/dppl_hmm_semisup.jl

@@ -1,3 +1,4 @@
+using FillArrays
 using StatsFuns: logsumexp
 
 # Set up hyperparameters
@@ -28,8 +29,8 @@ for t = 2:T_unsup
 end
 
 @model function dppl_hmm_semisup(K, T, T_unsup, w, z, u, alpha, beta)
-    theta ~ product_distribution(fill(Dirichlet(alpha), K))
-    phi ~ product_distribution(fill(Dirichlet(beta), K))
+    theta ~ product_distribution(Fill(Dirichlet(alpha), K))
+    phi ~ product_distribution(Fill(Dirichlet(beta), K))
     for t = 1:T
         w[t] ~ Categorical(phi[:, z[t]])
     end

models/dppl_lda.jl

@@ -1,3 +1,5 @@
+using FillArrays
+
 v = 100      # words
 k = 5        # topics
 m = 10       # number of docs
@@ -21,8 +23,8 @@ for i = 1:m
 end
 
 @model function dppl_lda(k, m, w, doc, alpha, beta)
-    theta ~ product_distribution(fill(Dirichlet(alpha), m))
-    phi ~ product_distribution(fill(Dirichlet(beta), k))
+    theta ~ product_distribution(Fill(Dirichlet(alpha), m))
+    phi ~ product_distribution(Fill(Dirichlet(beta), k))
     log_phi_dot_theta = log.(phi * theta)
     @addlogprob! sum(log_phi_dot_theta[CartesianIndex.(w, doc)])
 end

models/dppl_naive_bayes.jl

@@ -1,3 +1,4 @@
+using FillArrays
 using MLDatasets: MNIST
 using MultivariateStats: fit, PCA, transform
 
@@ -20,7 +21,7 @@ image_vec = image_subset[:, :]
 labels = labels[1:N]
 
 @model function dppl_naive_bayes(image_vec, labels, C, D)
-    m ~ product_distribution(fill(Normal(0, 10), C, D))
+    m ~ product_distribution(Fill(Normal(0, 10), C, D))
     image_vec ~ product_distribution(Normal.(m[labels, :]))
 end