Change getlogjoint -> getlogjoint_internal where needed

penelopeysm · penelopeysm · commit d7a46e1d7300 · 2025-07-26T18:06:23.000+01:00
diff --git a/ext/TuringDynamicHMCExt.jl b/ext/TuringDynamicHMCExt.jl
@@ -63,7 +63,7 @@ function DynamicPPL.initialstep(
 
     # Define log-density function.
     ℓ = DynamicPPL.LogDensityFunction(
-        model, DynamicPPL.getlogjoint, vi; adtype=spl.alg.adtype
+        model, DynamicPPL.getlogjoint_internal, vi; adtype=spl.alg.adtype
     )
 
     # Perform initial step.
diff --git a/src/mcmc/Inference.jl b/src/mcmc/Inference.jl
@@ -19,6 +19,7 @@ using DynamicPPL:
     setlogp!!,
     getlogp,
     getlogjoint,
+    getlogjoint_internal,
     VarName,
     getsym,
     getdist,
@@ -136,11 +137,13 @@ end
 
 Transition(θ, lp) = Transition(θ, lp, nothing)
 function Transition(model::DynamicPPL.Model, vi::AbstractVarInfo, t)
+    # TODO(DPPL0.37/penelopeysm): Fix this
     θ = getparams(model, vi)
-    lp = getlogjoint(vi)
+    lp = getlogjoint_internal(vi)
     return Transition(θ, lp, getstats(t))
 end
 
+# TODO(DPPL0.37/penelopeysm): Add log-prior and log-likelihood terms as well
 function metadata(t::Transition)
     stat = t.stat
     if stat === nothing
@@ -150,9 +153,11 @@ function metadata(t::Transition)
     end
 end
 
+# TODO(DPPL0.37/penelopeysm): Fix this
 DynamicPPL.getlogjoint(t::Transition) = t.lp
 
 # Metadata of VarInfo object
+# TODO(DPPL0.37/penelopeysm): Add log-prior and log-likelihood terms as well
 metadata(vi::AbstractVarInfo) = (lp=getlogjoint(vi),)
 
 ##########################
diff --git a/src/mcmc/emcee.jl b/src/mcmc/emcee.jl
@@ -72,7 +72,7 @@ function AbstractMCMC.step(
         vis[1],
         map(vis) do vi
             vi = DynamicPPL.link!!(vi, model)
-            AMH.Transition(vi[:], DynamicPPL.getlogjoint(vi), false)
+            AMH.Transition(vi[:], DynamicPPL.getlogjoint_internal(vi), false)
         end,
     )
 
@@ -87,7 +87,7 @@ function AbstractMCMC.step(
     densitymodel = AMH.DensityModel(
         Base.Fix1(
             LogDensityProblems.logdensity,
-            DynamicPPL.LogDensityFunction(model, DynamicPPL.getlogjoint, vi),
+            DynamicPPL.LogDensityFunction(model, DynamicPPL.getlogjoint_internal, vi),
         ),
     )
 
diff --git a/src/mcmc/external_sampler.jl b/src/mcmc/external_sampler.jl
@@ -163,7 +163,7 @@ function AbstractMCMC.step(
 
     # Construct LogDensityFunction
     f = DynamicPPL.LogDensityFunction(
-        model, DynamicPPL.getlogjoint, varinfo; adtype=alg.adtype
+        model, DynamicPPL.getlogjoint_internal, varinfo; adtype=alg.adtype
     )
 
     # Then just call `AbstractMCMC.step` with the right arguments.
diff --git a/src/mcmc/gibbs.jl b/src/mcmc/gibbs.jl
@@ -568,7 +568,7 @@ function setparams_varinfo!!(
     params::AbstractVarInfo,
 )
     logdensity = DynamicPPL.LogDensityFunction(
-        model, DynamicPPL.getlogjoint, state.ldf.varinfo; adtype=sampler.alg.adtype
+        model, DynamicPPL.getlogjoint_internal, state.ldf.varinfo; adtype=sampler.alg.adtype
     )
     new_inner_state = setparams_varinfo!!(
         AbstractMCMC.LogDensityModel(logdensity), sampler, state.state, params
diff --git a/src/mcmc/hmc.jl b/src/mcmc/hmc.jl
@@ -193,7 +193,7 @@ function DynamicPPL.initialstep(
     metricT = getmetricT(spl.alg)
     metric = metricT(length(theta))
     ldf = DynamicPPL.LogDensityFunction(
-        model, DynamicPPL.getlogjoint, vi; adtype=spl.alg.adtype
+        model, DynamicPPL.getlogjoint_internal, vi; adtype=spl.alg.adtype
     )
     lp_func = Base.Fix1(LogDensityProblems.logdensity, ldf)
     lp_grad_func = Base.Fix1(LogDensityProblems.logdensity_and_gradient, ldf)
@@ -308,7 +308,7 @@ end
 function get_hamiltonian(model, spl, vi, state, n)
     metric = gen_metric(n, spl, state)
     ldf = DynamicPPL.LogDensityFunction(
-        model, DynamicPPL.getlogjoint, vi; adtype=spl.alg.adtype
+        model, DynamicPPL.getlogjoint_internal, vi; adtype=spl.alg.adtype
     )
     lp_func = Base.Fix1(LogDensityProblems.logdensity, ldf)
     lp_grad_func = Base.Fix1(LogDensityProblems.logdensity_and_gradient, ldf)
diff --git a/src/mcmc/mh.jl b/src/mcmc/mh.jl
@@ -304,7 +304,7 @@ function propose!!(
 
     # Create a sampler and the previous transition.
     mh_sampler = AMH.MetropolisHastings(dt)
-    prev_trans = AMH.Transition(vt, DynamicPPL.getlogjoint(vi), false)
+    prev_trans = AMH.Transition(vt, DynamicPPL.getlogjoint_internal(vi), false)
 
     # Make a new transition.
     spl_model = DynamicPPL.contextualize(
@@ -313,7 +313,7 @@ function propose!!(
     densitymodel = AMH.DensityModel(
         Base.Fix1(
             LogDensityProblems.logdensity,
-            DynamicPPL.LogDensityFunction(spl_model, DynamicPPL.getlogjoint, vi),
+            DynamicPPL.LogDensityFunction(spl_model, DynamicPPL.getlogjoint_internal, vi),
         ),
     )
     trans, _ = AbstractMCMC.step(rng, densitymodel, mh_sampler, prev_trans)
@@ -341,7 +341,7 @@ function propose!!(
 
     # Create a sampler and the previous transition.
     mh_sampler = AMH.MetropolisHastings(spl.alg.proposals)
-    prev_trans = AMH.Transition(vals, DynamicPPL.getlogjoint(vi), false)
+    prev_trans = AMH.Transition(vals, DynamicPPL.getlogjoint_internal(vi), false)
 
     # Make a new transition.
     spl_model = DynamicPPL.contextualize(
@@ -350,7 +350,7 @@ function propose!!(
     densitymodel = AMH.DensityModel(
         Base.Fix1(
             LogDensityProblems.logdensity,
-            DynamicPPL.LogDensityFunction(spl_model, DynamicPPL.getlogjoint, vi),
+            DynamicPPL.LogDensityFunction(spl_model, DynamicPPL.getlogjoint_internal, vi),
         ),
     )
     trans, _ = AbstractMCMC.step(rng, densitymodel, mh_sampler, prev_trans)
diff --git a/src/mcmc/particle_mcmc.jl b/src/mcmc/particle_mcmc.jl
@@ -147,7 +147,7 @@ end
 
 function SMCTransition(model::DynamicPPL.Model, vi::AbstractVarInfo, weight)
     theta = getparams(model, vi)
-    lp = DynamicPPL.getlogjoint(vi)
+    lp = DynamicPPL.getlogjoint_internal(vi)
     return SMCTransition(theta, lp, weight)
 end
 
@@ -323,7 +323,7 @@ varinfo(state::PGState) = state.vi
 
 function PGTransition(model::DynamicPPL.Model, vi::AbstractVarInfo, logevidence)
     theta = getparams(model, vi)
-    lp = DynamicPPL.getlogjoint(vi)
+    lp = DynamicPPL.getlogjoint_internal(vi)
     return PGTransition(theta, lp, logevidence)
 end
 
diff --git a/src/mcmc/sghmc.jl b/src/mcmc/sghmc.jl
@@ -200,7 +200,7 @@ end
 
 function SGLDTransition(model::DynamicPPL.Model, vi::AbstractVarInfo, stepsize)
     theta = getparams(model, vi)
-    lp = DynamicPPL.getlogjoint(vi)
+    lp = DynamicPPL.getlogjoint_internal(vi)
     return SGLDTransition(theta, lp, stepsize)
 end
 
diff --git a/test/mcmc/external_sampler.jl b/test/mcmc/external_sampler.jl
@@ -21,7 +21,7 @@ function initialize_nuts(model::DynamicPPL.Model)
 
     # Create a LogDensityFunction
     f = DynamicPPL.LogDensityFunction(
-        model, DynamicPPL.getlogjoint, linked_vi; adtype=Turing.DEFAULT_ADTYPE
+        model, DynamicPPL.getlogjoint_internal, linked_vi; adtype=Turing.DEFAULT_ADTYPE
     )
 
     # Choose parameter dimensionality and initial parameter value
diff --git a/test/optimisation/Optimisation.jl b/test/optimisation/Optimisation.jl
@@ -41,35 +41,27 @@ using Turing
         @testset "With ConditionContext" begin
             m1 = model1(x)
             m2 = model2() | (x=x,)
-            # Doesn't matter if we use getlogjoint or getlogjoint_without_jacobian since the
+            # Doesn't matter if we use getlogjoint or getlogjoint_internal since the
             # VarInfo isn't linked.
-            ld1 = Turing.Optimisation.OptimLogDensity(
-                m1, Turing.Optimisation.getlogjoint_without_jacobian
-            )
-            ld2 = Turing.Optimisation.OptimLogDensity(m2, DynamicPPL.getlogjoint)
+            ld1 = Turing.Optimisation.OptimLogDensity(m1, DynamicPPL.getlogjoint)
+            ld2 = Turing.Optimisation.OptimLogDensity(m2, DynamicPPL.getlogjoint_internal)
             @test ld1(w) == ld2(w)
         end
 
         @testset "With prefixes" begin
             vn = @varname(inner)
             m1 = prefix(model1(x), vn)
             m2 = prefix((model2() | (x=x,)), vn)
-            ld1 = Turing.Optimisation.OptimLogDensity(
-                m1, Turing.Optimisation.getlogjoint_without_jacobian
-            )
-            ld2 = Turing.Optimisation.OptimLogDensity(m2, DynamicPPL.getlogjoint)
+            ld1 = Turing.Optimisation.OptimLogDensity(m1, DynamicPPL.getlogjoint)
+            ld2 = Turing.Optimisation.OptimLogDensity(m2, DynamicPPL.getlogjoint_internal)
             @test ld1(w) == ld2(w)
         end
 
         @testset "Joint, prior, and likelihood" begin
             m1 = model1(x)
             a = [0.3]
-            ld_joint = Turing.Optimisation.OptimLogDensity(
-                m1, Turing.Optimisation.getlogjoint_without_jacobian
-            )
-            ld_prior = Turing.Optimisation.OptimLogDensity(
-                m1, Turing.Optimisation.getlogprior_without_jacobian
-            )
+            ld_joint = Turing.Optimisation.OptimLogDensity(m1, DynamicPPL.getlogjoint)
+            ld_prior = Turing.Optimisation.OptimLogDensity(m1, DynamicPPL.getlogprior)
             ld_likelihood = Turing.Optimisation.OptimLogDensity(
                 m1, DynamicPPL.getloglikelihood
             )

Original file line number	Diff line number	Diff line change
`@@ -63,7 +63,7 @@ function DynamicPPL.initialstep(`
`63`	`63`
`64`	`64`	`# Define log-density function.`
`65`	`65`	`ℓ = DynamicPPL.LogDensityFunction(`
`66`		`- model, DynamicPPL.getlogjoint, vi; adtype=spl.alg.adtype`
	`66`	`+ model, DynamicPPL.getlogjoint_internal, vi; adtype=spl.alg.adtype`
`67`	`67`	`)`
`68`	`68`
`69`	`69`	`# Perform initial step.`
Original file line number	Diff line number	Diff line change
`@@ -163,7 +163,7 @@ function AbstractMCMC.step(`
`163`	`163`
`164`	`164`	`# Construct LogDensityFunction`
`165`	`165`	`f = DynamicPPL.LogDensityFunction(`
`166`		`- model, DynamicPPL.getlogjoint, varinfo; adtype=alg.adtype`
	`166`	`+ model, DynamicPPL.getlogjoint_internal, varinfo; adtype=alg.adtype`
`167`	`167`	`)`
`168`	`168`
`169`	`169`	# Then just call `AbstractMCMC.step` with the right arguments.
Original file line number	Diff line number	Diff line change
`@@ -568,7 +568,7 @@ function setparams_varinfo!!(`
`568`	`568`	`params::AbstractVarInfo,`
`569`	`569`	`)`
`570`	`570`	`logdensity = DynamicPPL.LogDensityFunction(`
`571`		`- model, DynamicPPL.getlogjoint, state.ldf.varinfo; adtype=sampler.alg.adtype`
	`571`	`+ model, DynamicPPL.getlogjoint_internal, state.ldf.varinfo; adtype=sampler.alg.adtype`
`572`	`572`	`)`
`573`	`573`	`new_inner_state = setparams_varinfo!!(`
`574`	`574`	`AbstractMCMC.LogDensityModel(logdensity), sampler, state.state, params`
Original file line number	Diff line number	Diff line change
`@@ -21,7 +21,7 @@ function initialize_nuts(model::DynamicPPL.Model)`
`21`	`21`
`22`	`22`	`# Create a LogDensityFunction`
`23`	`23`	`f = DynamicPPL.LogDensityFunction(`
`24`		`- model, DynamicPPL.getlogjoint, linked_vi; adtype=Turing.DEFAULT_ADTYPE`
	`24`	`+ model, DynamicPPL.getlogjoint_internal, linked_vi; adtype=Turing.DEFAULT_ADTYPE`
`25`	`25`	`)`
`26`	`26`
`27`	`27`	`# Choose parameter dimensionality and initial parameter value`