Use Random.default_rng() (#1967)

Project.toml

@@ -1,6 +1,6 @@
 name = "Turing"
 uuid = "fce5fe82-541a-59a6-adf8-730c64b5f9a0"
-version = "0.24.2"
+version = "0.24.3"
 
 [deps]
 AbstractMCMC = "80f14c24-f653-4e6a-9b94-39d6b0f70001"

docs/src/for-developers/compiler.md

@@ -30,7 +30,7 @@ A `model::Model` is a callable struct that one can sample from by calling
 ```julia
 (model::Model)([rng, varinfo, sampler, context])
 ```
-where `rng` is a random number generator (default: `Random.GLOBAL_RNG`), `varinfo` is a data structure that stores information
+where `rng` is a random number generator (default: `Random.default_rng()`), `varinfo` is a data structure that stores information
 about the random variables (default: `DynamicPPL.VarInfo()`), `sampler` is a sampling algorithm (default: `DynamicPPL.SampleFromPrior()`),
 and `context` is a sampling context that can, e.g., modify how the log probability is accumulated (default: `DynamicPPL.DefaultContext()`).
 

docs/src/using-turing/performancetips.md

@@ -116,7 +116,7 @@ model = tmodel(1.0)
     model,
     Turing.VarInfo(model),
     Turing.SamplingContext(
-        Random.GLOBAL_RNG, Turing.SampleFromPrior(), Turing.DefaultContext(),
+        Random.default_rng(), Turing.SampleFromPrior(), Turing.DefaultContext(),
     ),
     model.args...,
 )

src/inference/Inference.jl

@@ -133,7 +133,7 @@ function AbstractMCMC.sample(
     N::Integer;
     kwargs...
 )
-    return AbstractMCMC.sample(Random.GLOBAL_RNG, model, alg, N; kwargs...)
+    return AbstractMCMC.sample(Random.default_rng(), model, alg, N; kwargs...)
 end
 
 function AbstractMCMC.sample(
@@ -190,7 +190,7 @@ function AbstractMCMC.sample(
     n_chains::Integer;
     kwargs...
 )
-    return AbstractMCMC.sample(Random.GLOBAL_RNG, model, alg, ensemble, N, n_chains;
+    return AbstractMCMC.sample(Random.default_rng(), model, alg, ensemble, N, n_chains;
                                kwargs...)
 end
 
@@ -397,7 +397,7 @@ function save(c::MCMCChains.Chains, spl::Sampler, model, vi, samples)
 end
 
 function resume(chain::MCMCChains.Chains, args...; kwargs...)
-    return resume(Random.GLOBAL_RNG, chain, args...; kwargs...)
+    return resume(Random.default_rng(), chain, args...; kwargs...)
 end
 
 function resume(rng::Random.AbstractRNG, chain::MCMCChains.Chains, args...;
@@ -530,7 +530,7 @@ true
 ```
 """
 function predict(model::Model, chain::MCMCChains.Chains; kwargs...)
-    return predict(Random.GLOBAL_RNG, model, chain; kwargs...)
+    return predict(Random.default_rng(), model, chain; kwargs...)
 end
 function predict(rng::AbstractRNG, model::Model, chain::MCMCChains.Chains; include_all = false)
     # Don't need all the diagnostics
@@ -617,7 +617,7 @@ function transitions_from_chain(
     chain::MCMCChains.Chains;
     kwargs...
 )
-    return transitions_from_chain(Random.GLOBAL_RNG, model, chain; kwargs...)
+    return transitions_from_chain(Random.default_rng(), model, chain; kwargs...)
 end
 
 function transitions_from_chain(

src/variational/advi.jl

@@ -91,7 +91,7 @@ end
 
 Creates a mean-field approximation with multivariate normal as underlying distribution.
 """
-meanfield(model::DynamicPPL.Model) = meanfield(Random.GLOBAL_RNG, model)
+meanfield(model::DynamicPPL.Model) = meanfield(Random.default_rng(), model)
 function meanfield(rng::Random.AbstractRNG, model::DynamicPPL.Model)
     # Setup.
     varinfo = DynamicPPL.VarInfo(model)

test/inference/gibbs.jl

@@ -21,7 +21,7 @@
         # Test gid of each samplers
         g = Turing.Sampler(s3, gdemo_default)
 
-        _, state = AbstractMCMC.step(Random.GLOBAL_RNG, gdemo_default, g)
+        _, state = AbstractMCMC.step(Random.default_rng(), gdemo_default, g)
         @test state.samplers[1].selector != g.selector
         @test state.samplers[2].selector != g.selector
         @test state.samplers[1].selector != state.samplers[2].selector