Merge remote-tracking branch 'origin/main' into breaking

Author: penelopeysm

HISTORY.md

@@ -70,6 +70,10 @@ And a couple of more internal changes:
   - The model evaluation function, `model.f` for some `model::Model`, no longer takes a context as an argument
   - The internal representation and API dealing with submodels (i.e., `ReturnedModelWrapper`, `Sampleable`, `should_auto_prefix`, `is_rhs_model`) has been simplified. If you need to check whether something is a submodel, just use `x isa DynamicPPL.Submodel`. Note that the public API i.e. `to_submodel` remains completely untouched.
 
+## 0.36.15
+
+Bumped minimum Julia version to 1.10.8 to avoid potential crashes with `Core.Compiler.widenconst` (which Mooncake uses).
+
 ## 0.36.14
 
 Added compatibility with AbstractPPL@0.12.

Project.toml

@@ -74,4 +74,4 @@ Random = "1.6"
 Requires = "1"
 Statistics = "1"
 Test = "1.6"
-julia = "1.10"
+julia = "1.10.8"

src/test_utils/models.jl

@@ -14,12 +14,12 @@ A model with variables `m` and `x` with `x` having support depending on `m`.
 """
 @model function demo_dynamic_constraint()
     m ~ Normal()
-    x ~ truncated(Normal(), m, Inf)
+    x ~ truncated(Normal(); lower=m)
 
     return (m=m, x=x)
 end
 function logprior_true(model::Model{typeof(demo_dynamic_constraint)}, m, x)
-    return logpdf(Normal(), m) + logpdf(truncated(Normal(), m, Inf), x)
+    return logpdf(Normal(), m) + logpdf(truncated(Normal(); lower=m), x)
 end
 function loglikelihood_true(model::Model{typeof(demo_dynamic_constraint)}, m, x)
     return zero(float(eltype(m)))
@@ -30,7 +30,7 @@ end
 function logprior_true_with_logabsdet_jacobian(
     model::Model{typeof(demo_dynamic_constraint)}, m, x
 )
-    b_x = Bijectors.bijector(truncated(Normal(), m, Inf))
+    b_x = Bijectors.bijector(truncated(Normal(); lower=m))
     x_unconstrained, Δlogp = Bijectors.with_logabsdet_jacobian(b_x, x)
     return (m=m, x=x_unconstrained), logprior_true(model, m, x) - Δlogp
 end

test/compiler.jl

@@ -498,7 +498,7 @@ module Issue537 end
         @model function demo(y)
             α ~ Uniform()
             μ ~ Normal()
-            σ ~ truncated(Normal(), 0, Inf)
+            σ ~ truncated(Normal(); lower=0)
             num_steps = length(y[1])
             num_obs = length(y)
             @inbounds for i in 1:num_obs

test/varinfo.jl

@@ -67,16 +67,8 @@ end
 
     @testset "Base" begin
         # Test Base functions:
-        #   string, Symbol, ==, hash, in, keys, haskey, isempty, push!!, empty!!,
+        #   in, keys, haskey, isempty, push!!, empty!!,
         #   getindex, setindex!, getproperty, setproperty!
-        csym = gensym()
-        vn1 = @varname x[1][2]
-        @test string(vn1) == "x[1][2]"
-        @test Symbol(vn1) == Symbol("x[1][2]")
-
-        vn2 = @varname x[1][2]
-        @test vn2 == vn1
-        @test hash(vn2) == hash(vn1)
 
         function test_base(vi_original)
             vi = deepcopy(vi_original)
@@ -289,14 +281,14 @@ end
     @testset "setval! & setval_and_resample!" begin
         @model function testmodel(x)
             n = length(x)
-            s ~ truncated(Normal(), 0, Inf)
+            s ~ truncated(Normal(); lower=0)
             m ~ MvNormal(zeros(n), I)
             return x ~ MvNormal(m, s^2 * I)
         end
 
         @model function testmodel_univariate(x, ::Type{TV}=Vector{Float64}) where {TV}
             n = length(x)
-            s ~ truncated(Normal(), 0, Inf)
+            s ~ truncated(Normal(); lower=0)
 
             m = TV(undef, n)
             for i in eachindex(m)
@@ -561,10 +553,10 @@ end
     end
 
     @testset "istrans" begin
-        @model demo_constrained() = x ~ truncated(Normal(), 0, Inf)
+        @model demo_constrained() = x ~ truncated(Normal(); lower=0)
         model = demo_constrained()
         vn = @varname(x)
-        dist = truncated(Normal(), 0, Inf)
+        dist = truncated(Normal(); lower=0)
 
         ### `VarInfo`
         # Need to run once since we can't specify that we want to _sample_