Dev (#309)

* update Project.toml

* s/logpdf/logdensity/

* update deps

* scratchpad

* version bump

* representative => rootmeasure

* update dependency versions

* reduce dependencies

* update deps

* scratchpad

* representative => rootmeasure

* update dependency versions

* reduce dependencies

* `as` methods for `xform`

* cleanup

* require latest MeasureTheory

* dorp old distributions code

* drop old iid code

* drop extra space

* limit deps to three newest releases

* update dynamichmc

* add Aqua

* bump version

* Better `predict` method

* withmeasures(::ConditionalModel)

* update dependencies

* updating symbolics

* some updates to symbolics

* hmm example

* update MeasureBase bound

* better dispatch for `predict`

* drop redundant method

* remove whitespace

* bump version

Project.toml

@@ -1,7 +1,7 @@
 name = "Soss"
 uuid = "8ce77f84-9b61-11e8-39ff-d17a774bf41c"
 author = ["Chad Scherrer <chad.scherrer@gmail.com>"]
-version = "0.20.8"
+version = "0.20.9"
 
 [deps]
 ArrayInterface = "4fba245c-0d91-5ea0-9b3e-6abc04ee57a9"
@@ -49,7 +49,7 @@ JuliaVariables = "0.2"
 MLStyle = "0.3,0.4"
 MacroTools = "0.5"
 MappedArrays = "0.3, 0.4"
-MeasureBase = "0.4"
+MeasureBase = "0.5"
 MeasureTheory = "0.13"
 NamedTupleTools = "0.12, 0.13"
 NestedTuples = "0.3"
@@ -65,7 +65,7 @@ SpecialFunctions = "0.9, 0.10, 1"
 StatsBase = "0.33"
 StatsFuns = "0.9"
 SymbolicCodegen = "0.2"
-SymbolicUtils = "0.14, 0.15, 0.16"
+SymbolicUtils = "0.15, 0.16, 0.17"
 TransformVariables = "0.4"
 TupleVectors = "0.1"
 julia = "1.5"

scratchpad/hmm.jl

@@ -0,0 +1,48 @@
+using MeasureTheory
+using Base.Iterators
+using Statistics
+
+using Random
+rng = Random.Xoshiro(3)
+
+x = Chain(Normal()) do xj Normal(μ=xj) end
+xobs = rand(rng, x)
+y = For(xobs) do xj Poisson(logλ=xj) end
+yobs = rand(rng, y)
+xv = take(xobs, 10) |> collect
+yv = take(yobs, 10) |> collect
+
+take(xobs.parent, 10) |> collect
+take(yobs.parent, 10) |> collect
+
+
+exp.(xv)
+yv
+
+# using Plots
+
+# plt = scatter(normcdf.(xv, 1, yv), label=false)
+
+# for j in 1:10
+#     xobs = rand(rng, x)
+#     yobs = rand(rng, y)
+#     xv = take(xobs, 100) |> collect;
+#     yv = take(yobs, 100) |> collect;
+#     plt = scatter!(plt, normcdf.(xv, 1, yv), label=false)
+# end
+# plt
+
+using Soss
+
+m = @model begin
+    x ~ Chain(Normal()) do xj Normal(μ=xj) end
+    y ~ For(xobs) do xj Poisson(logλ=xj) end
+end
+
+truth = rand(rng, m())
+
+xobs = take(truth.x, 10) |> collect
+yobs = take(truth.y, 10) |> collect
+
+logdensity(m(), (x=xobs, y=yobs))
+

src/symbolic/codegen.jl

@@ -32,6 +32,7 @@ function SymbolicCodegen.codegen(cm :: ConditionalModel; kwargs...)
         pushfirst!(code.args, :($v = getproperty(_pars, $vname)))
     end
 
+    code = MacroTools.flatten(code)
 
     return mk_function(getmodule(cm), (:_args, :_data, :_pars), (), code)
 

src/symbolic/symbolic.jl

@@ -19,17 +19,17 @@ export schema
 
 export symlogdensity
 
-symlogdensity(d, x::Symbolic) = logdensity(d,x)
+symlogdensity(d, x::Symbolic) = logpdf(d,x)
 
-function symlogdensity(d::ProductMeasure{<:AbstractArray}, x::Symbolic{A}) where {A <: AbstractArray}
+function symlogdensity(d::ProductMeasure{F,S,<:AbstractArray}, x::Symbolic{A}) where {F,S,A <: AbstractArray}
     dims = size(d)
 
     iters = Sym{Int}.(gensym.(Symbol.(:i, 1:length(dims))))
 
-    marginals = d.data
+    mar = marginals(d)
 
     # To begin, the result is just the summand
-    result = getsummand(marginals, x, iters)
+    result = getsummand(mar, x, iters)
 
     # Then we wrap in a summation index for each dimension
     for i in 1:length(dims)

src/transforms/predict.jl

@@ -1,10 +1,13 @@
 export predict
 using TupleVectors
+using SampleChains
 
-predict(m::AbstractModel, args...) = predict(Random.GLOBAL_RNG, m, args...)
-predict(d::AbstractMeasure, x) = x
+predict(args...; kwargs...) = predict(Random.GLOBAL_RNG, args...; kwargs...)
 
+# TODO: Fix this hack
+predict(d::AbstractMeasure, x) = x
 predict(d::Dists.Distribution, x) = x
+predict(d::AbstractModel, args...; kwargs...) = predict(Random.GLOBAL_RNG, d, args...; kwargs...)
 
 @inline function predict(rng::AbstractRNG, m::AbstractModel, nt::NamedTuple{N}) where {N}
     pred = predictive(Model(m), N...)
@@ -13,7 +16,6 @@ end
 
 predict(rng::AbstractRNG, m::AbstractModel; kwargs...) = predict(rng, m, (;kwargs...))
 
-
 @inline function predict(rng::AbstractRNG, d::AbstractModel, nt::LazyMerge)
     predict(rng, d, convert(NamedTuple, nt))
 end
@@ -33,8 +35,6 @@ function predict(rng::AbstractRNG, d::AbstractModel, post::AbstractVector{<:Name
     v
 end
 
-using SampleChains
-
 function predict(rng::AbstractRNG, d::ConditionalModel, post::MultiChain)
     [predict(rng, d, c) for c in getchains(post)]
 end