Remove Optim.jl extension

Author: penelopeysm

Project.toml

@@ -41,11 +41,9 @@ StatsFuns = "4c63d2b9-4356-54db-8cca-17b64c39e42c"
 
 [weakdeps]
 DynamicHMC = "bbc10e6e-7c05-544b-b16e-64fede858acb"
-Optim = "429524aa-4258-5aef-a3af-852621145aeb"
 
 [extensions]
 TuringDynamicHMCExt = "DynamicHMC"
-TuringOptimExt = ["Optim", "AbstractPPL"]
 
 [compat]
 ADTypes = "1.9"
@@ -72,7 +70,6 @@ LinearAlgebra = "1"
 LogDensityProblems = "2"
 MCMCChains = "5, 6, 7"
 NamedArrays = "0.9, 0.10"
-Optim = "1"
 Optimization = "3, 4, 5"
 OptimizationOptimJL = "0.1, 0.2, 0.3, 0.4"
 OrderedCollections = "1"
@@ -86,7 +83,3 @@ StatsAPI = "1.6"
 StatsBase = "0.32, 0.33, 0.34"
 StatsFuns = "0.8, 0.9, 1"
 julia = "1.10.8"
-
-[extras]
-DynamicHMC = "bbc10e6e-7c05-544b-b16e-64fede858acb"
-Optim = "429524aa-4258-5aef-a3af-852621145aeb"

ext/TuringOptimExt.jl

@@ -19,28 +19,26 @@ using Statistics: Statistics
 using LinearAlgebra: LinearAlgebra
 
 export maximum_a_posteriori, maximum_likelihood
-# The MAP and MLE exports are only needed for the Optim.jl interface.
-export MAP, MLE
 
 """
     ModeEstimator
 
 An abstract type to mark whether mode estimation is to be done with maximum a posteriori
-(MAP) or maximum likelihood estimation (MLE). This is only needed for the Optim.jl interface.
+(MAP) or maximum likelihood estimation (MLE).
 """
 abstract type ModeEstimator end
 
 """
     MLE <: ModeEstimator
 
-Concrete type for maximum likelihood estimation. Only used for the Optim.jl interface.
+Concrete type for maximum likelihood estimation.
 """
 struct MLE <: ModeEstimator end
 
 """
     MAP <: ModeEstimator
 
-Concrete type for maximum a posteriori estimation. Only used for the Optim.jl interface.
+Concrete type for maximum a posteriori estimation.
 """
 struct MAP <: ModeEstimator end
 
@@ -123,32 +121,6 @@ required by Optimization.jl.
 (f::OptimLogDensity)(z::AbstractVector) = -LogDensityProblems.logdensity(f.ldf, z)
 (f::OptimLogDensity)(z, _) = f(z)
 
-# NOTE: The format of this function is dictated by Optim. The first argument sets whether to
-# compute the function value, the second whether to compute the gradient (and stores the
-# gradient). The last one is the actual argument of the objective function.
-function (f::OptimLogDensity)(F, G, z)
-    if G !== nothing
-        # Calculate log joint and its gradient.
-        logp, ∇logp = LogDensityProblems.logdensity_and_gradient(f.ldf, z)
-
-        # Save the negative gradient to the pre-allocated array.
-        copyto!(G, -∇logp)
-
-        # If F is something, the negative log joint is requested as well.
-        # We have already computed it as a by-product above and hence return it directly.
-        if F !== nothing
-            return -logp
-        end
-    end
-
-    # Only negative log joint requested but no gradient.
-    if F !== nothing
-        return -LogDensityProblems.logdensity(f.ldf, z)
-    end
-
-    return nothing
-end
-
 """
     ModeResult{
         V<:NamedArrays.NamedArray,