Some clean up + making MALA available without ForwardDiff (#78)

* removed unnecessary unions and double definitions

* make MALA available even when ForwardDiff isn't exportedc

* forgot to add src/forwarddiff.jl

* version bump

* fixed method ambiguity

* fix constructor for MALA

* added check for capabilties of LogDensityModel

Author: torfjelde

Project.toml

@@ -1,6 +1,6 @@
 name = "AdvancedMH"
 uuid = "5b7e9947-ddc0-4b3f-9b55-0d8042f74170"
-version = "0.7.1"
+version = "0.7.2"
 
 [deps]
 AbstractMCMC = "80f14c24-f653-4e6a-9b94-39d6b0f70001"

src/AdvancedMH.jl

@@ -66,16 +66,15 @@ function Transition(model::AbstractMCMC.LogDensityModel, params)
 end
 
 # Calculate the density of the model given some parameterization.
-logdensity(model::DensityModelOrLogDensityModel, params) = model.logdensity(params)
-logdensity(model::DensityModelOrLogDensityModel, t::Transition) = t.lp
-
+logdensity(model::DensityModel, params) = model.logdensity(params)
+logdensity(::DensityModel, t::Transition) = t.lp
 logdensity(model::AbstractMCMC.LogDensityModel, params) = LogDensityProblems.logdensity(model.logdensity, params)
-logdensity(model::AbstractMCMC.LogDensityModel, t::Transition) = t.lp
+logdensity(::AbstractMCMC.LogDensityModel, t::Transition) = t.lp
 
 # A basic chains constructor that works with the Transition struct we defined.
 function AbstractMCMC.bundle_samples(
     ts::Vector{<:AbstractTransition},
-    model::Union{<:DensityModelOrLogDensityModel,<:AbstractMCMC.LogDensityModel},
+    model::DensityModelOrLogDensityModel,
     sampler::MHSampler,
     state,
     chain_type::Type{Vector{NamedTuple}};
@@ -101,7 +100,7 @@ end
 
 function AbstractMCMC.bundle_samples(
     ts::Vector{<:Transition{<:NamedTuple}},
-    model::Union{<:DensityModelOrLogDensityModel,<:AbstractMCMC.LogDensityModel},
+    model::DensityModelOrLogDensityModel,
     sampler::MHSampler,
     state,
     chain_type::Type{Vector{NamedTuple}};
@@ -122,13 +121,14 @@ function __init__()
     @require MCMCChains="c7f686f2-ff18-58e9-bc7b-31028e88f75d" include("mcmcchains-connect.jl")
     @require StructArrays="09ab397b-f2b6-538f-b94a-2f83cf4a842a" include("structarray-connect.jl")
     @require DiffResults = "163ba53b-c6d8-5494-b064-1a9d43ac40c5" begin
-        @require ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210" include("MALA.jl")
+        @require ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210" include("forwarddiff.jl")
     end
 end
 
 # Include inference methods.
 include("proposal.jl")
 include("mh-core.jl")
 include("emcee.jl")
+include("MALA.jl")
 
 end # module AdvancedMH

src/MALA.jl

@@ -1,17 +1,15 @@
-using .ForwardDiff: gradient!
-using .DiffResults: GradientResult, value, gradient
-
 struct MALA{D} <: MHSampler
     proposal::D
+
+    MALA{D}(proposal::D) where {D} = new{D}(proposal)
 end
 
+# If we were given a RandomWalkProposal, just use that instead.
+MALA(d::RandomWalkProposal) = MALA{typeof(d)}(d)
 
 # Create a RandomWalkProposal if we weren't given one already.
 MALA(d) = MALA(RandomWalkProposal(d))
 
-# If we were given a RandomWalkProposal, just use that instead.
-MALA(d::RandomWalkProposal) = MALA{typeof(d)}(d)
-
 
 struct GradientTransition{T<:Union{Vector, Real, NamedTuple}, L<:Real, G<:Union{Vector, Real, NamedTuple}} <: AbstractTransition
     params::T
@@ -85,9 +83,7 @@ end
 Return the value and gradient of the log density of the parameters `params` for the `model`.
 """
 function logdensity_and_gradient(model::DensityModel, params)
-    res = GradientResult(params)
-    gradient!(res, model.logdensity, params)
-    return value(res), gradient(res)
+    error("no implementation exist for `DensityModel`; try importing ForwardDiff.jl")
 end
 
 """
@@ -96,7 +92,8 @@ end
 Return the value and gradient of the log density of the parameters `params` for the `model`.
 """
 function logdensity_and_gradient(model::AbstractMCMC.LogDensityModel, params)
-     return LogDensityProblems.logdensity_and_gradient(model.logdensity, params)
+    check_capabilities(model)
+    return LogDensityProblems.logdensity_and_gradient(model.logdensity, params)
  end
 
 

src/forwarddiff.jl

@@ -0,0 +1,9 @@
+using .ForwardDiff: gradient!
+using .DiffResults: GradientResult, value, gradient
+using .AdvancedMH: AdvancedMH
+
+function AdvancedMH.logdensity_and_gradient(model::AdvancedMH.DensityModel, params)
+    res = GradientResult(params)
+    gradient!(res, model.logdensity, params)
+    return value(res), gradient(res)
+end