added extension to make it work with Turing.jl

Project.toml

@@ -1,6 +1,6 @@
 name = "AutomaticMALA"
 uuid = "fa450ca5-f5a1-411d-b5db-2b752c7e65ee"
-authors = ["Tor Erlend Fjelde <tor.erlend95@gmail.com> and contributors"]
+authors = ["Tor Erlend Fjelde <tor.github@gmail.com> and contributors"]
 version = "0.1.0"
 
 [deps]
@@ -11,5 +11,19 @@ LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 LogDensityProblems = "6fdf6af0-433a-55f7-b3ed-c6c6e0b8df7c"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 
+[weakdeps]
+Turing = "fce5fe82-541a-59a6-adf8-730c64b5f9a0"
+
+[extensions]
+AutomaticMALATuringExt = ["Turing"]
+
 [compat]
+AbstractMCMC = "5"
+Distributions = "0.25"
+DocStringExtensions = "0.9"
+LogDensityProblems = "2"
+Turing = "0.31"
 julia = "1.6"
+
+[extras]
+Turing = "fce5fe82-541a-59a6-adf8-730c64b5f9a0"

ext/AutomaticMALATuringExt.jl

@@ -0,0 +1,13 @@
+module AutomaticMALATuringExt
+
+if isdefined(Base, :get_extension)
+    using AutomaticMALA: AutomaticMALA
+    using Turing: Turing
+else
+    using ..AutomaticMALA: AutomaticMALA
+    using ..Turing: Turing
+end
+
+Turing.Inference.getparams(::Turing.DynamicPPL.Model, state::AutomaticMALA.AutoMALAState) = state.x
+
+end

src/AutomaticMALA.jl

@@ -210,4 +210,16 @@ function round_based_adaptation(
     return AutoMALA(ϵ_init, 0), initial_params
 end
 
+if !isdefined(Base, :get_extension)
+    using Requires
+end
+
+@static if !isdefined(Base, :get_extension)
+    function __init__()
+        @require Turing = "fce5fe82-541a-59a6-adf8-730c64b5f9a0" include(
+            "../ext/AutomaticMALATuringExt.jl"
+        )
+    end
+end
+
 end