Dev (#193)

* Dirichlet(k::Integer, α) = Dirichlet(Fill(α, k))

* export TransformVariables as TV

* drop redundant import

* 0.0 => zero(Float64)

* drop outdated Dists.logpdf

* update StudentT

* drop redundant import

* update Uniform

* bump MeasureBase version

* reworking beta

* small update to StudentT

* basemeasure for discrete Distributions

* using LogExpFunctions => import LogExpFunctions

* quoteof(::Chain)

* prettyprinting and chain-mucking

* Some refactoring for Markov chains

* import MeasureBase: ≪

* version bound for PrettyPrinting

* copy(rng) might change its type (e.g. GLOBAL_RNG)

* tests pass

* cleaning up

* more cleanup

* big update

* get tests passing

* formatting

* oops typo

* move affine to MeasureTheory

* updating

* Val => StaticSymbol

* more fixes

* fix fix fix

* more logdesnity => logdensity_def

* more logdesnity fixes

* debugging

* formatting

* bugfixes

* working on tests

* updates

* working on tests

* tests passing!

* refactor

* working on tests

* drop static weight for now

* fix sampling from ProductMeasure{<:Base.Generator}

* tests passing!!

* more stuff

* constructor => constructorof

* constructor =? construtorof

* updates

* working on tests

* fix Dirichlet

* update Bernoulli

* working on tests

* bugfixes for RealizedSamples

* tests passing!!

* tighten down inference

* as(::PowerMeasure)

* drop type-level stuff

* using InverseFunctions.jl

* update license

* affero

* copyright

* update CI to 1.6

* xform => TV.as

* oops missed a conflict

* fix merge corruption

* typo

* fix license

* Update README.md

* merge

* enumerate instead of zip

* bugfix

* inline rand

* drop `static` from `insupport` results

* update proxies

* Move ConditionalMeasure to MeasureBase

* IfElse.ifelse(p::Bernoulli, t, f)

* IfElseMeasure

* update some base measures

* test broken :(

* fix some redundancies

* instance_type => Core.Typeof

* update testvalue for Bernoulli and Binomial

* un-break broken test (now passing)

* Fall-back `For` method for when inference fails

* drop extra spaces

* more whitespace

* bump MeasureBase dependency version

* add newline

* tidy up

* ifelse tests

* OEF newline

* avoid type piracy

* add Julia 1.7 to CI

* make Julia 1.6 happy

* approx instead of ==

* Require at least Julia 1.6

* Try Sebastian's idea test_measures ::Any[]

* Another Any[]

* Drop Likelihood test

* drop 1.7 CI (seems buggy?)

* bump version

* export likelihood

* Snedecor's F

* Gamma distribution

* more gamma stuff

* Beroulli()

* inverse Gaussian

* Getting modifed GLM.jl tests to pass

* drop pdf and logpdf

* Poisson bugfix

* update Normal(μ, σ²)

* Gamma(μ, ϕ) for GLMs

* updates for GLM support

* start on truncated

* update parameterized measures

* drop FactoredBase

* drop old LazyArrays dependency

* insupport(::Distribution)

* Left out"Dists."

* don't export `ifelse` (#192)

* Kleisli => TransitionKernel

* depend on StatsBase

* tests passing

* bump MeasureBase version

* work on truncated and censored

* improve func_string

* Simplify logdensity_def(::For, x)

* Move truncated and censored updates to separate branches

* newline

* comment out in-progress stuff

* newline

* bump version

* update formatting spec

* more formatting

* tweedie docs

* drop redundant exports

* update exports

* omega => lambda

* drop SequentialEx

* get tests passing

* add kernel tests

* gitignore

* better `Pretty.tile` for Affine and AffineTransforms

* formatting

* kleisli => kernel

* update tile(::For)

* update Compat version

* bump MB version

* update gamma

* Let's come back to InverseGaussian

* CI on 1.7

* update IfElse

* formatting

.JuliaFormatter.toml

@@ -6,7 +6,7 @@ whitespace_ops_in_indices = false
 remove_extra_newlines = true
 import_to_using = false
 pipe_to_function_call = false
-short_to_long_function_def = false
+short_to_long_function_def = true
 always_use_return = false
 whitespace_in_kwargs = true
 annotate_untyped_fields_with_any = false

.github/workflows/ci.yml

@@ -23,6 +23,7 @@ jobs:
       matrix:
         version:
           - '1.6'
+          - '1.7'
         os:
           - ubuntu-latest
           - macOS-latest

.gitignore

@@ -9,3 +9,11 @@
 /.vscode/
 /coverage
 coverage-lcov.info
+paper/paper.aux
+paper/paper.bbl
+paper/paper.blg
+paper/paper.fdb_latexmk
+paper/paper.fls
+paper/paper.log
+paper/paper.out
+paper/paper.synctex.gz

Project.toml

@@ -1,7 +1,7 @@
 name = "MeasureTheory"
 uuid = "eadaa1a4-d27c-401d-8699-e962e1bbc33b"
 authors = ["Chad Scherrer <[email protected]> and contributors"]
-version = "0.15.1"
+version = "0.16.0"
 
 [deps]
 Accessors = "7d9f7c33-5ae7-4f3b-8dc6-eff91059b697"
@@ -12,14 +12,12 @@ ConstructionBase = "187b0558-2788-49d3-abe0-74a17ed4e7c9"
 DensityInterface = "b429d917-457f-4dbc-8f4c-0cc954292b1d"
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 DynamicIterators = "6c76993d-992e-5bf1-9e63-34920a5a5a38"
-FLoops = "cc61a311-1640-44b5-9fba-1b764f453329"
 FillArrays = "1a297f60-69ca-5386-bcde-b61e274b549b"
 IfElse = "615f187c-cbe4-4ef1-ba3b-2fcf58d6d173"
 Infinities = "e1ba4f0e-776d-440f-acd9-e1d2e9742647"
 InteractiveUtils = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
 InverseFunctions = "3587e190-3f89-42d0-90ee-14403ec27112"
 KeywordCalls = "4d827475-d3e4-43d6-abe3-9688362ede9f"
-LazyArrays = "5078a376-72f3-5289-bfd5-ec5146d43c02"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 LogExpFunctions = "2ab3a3ac-af41-5b50-aa03-7779005ae688"
 MLStyle = "d8e11817-5142-5d16-987a-aa16d5891078"
@@ -35,31 +33,31 @@ Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b"
 Static = "aedffcd0-7271-4cad-89d0-dc628f76c6d3"
 StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
+Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
+StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 StatsFuns = "4c63d2b9-4356-54db-8cca-17b64c39e42c"
 TransformVariables = "84d833dd-6860-57f9-a1a7-6da5db126cff"
 Tricks = "410a4b4d-49e4-4fbc-ab6d-cb71b17b3775"
 
 [compat]
 Accessors = "0.1"
 ChangesOfVariables = "0.1"
-Compat = "3.35"
+Compat = "3.35, 4"
 ConcreteStructs = "0.2"
 ConstructionBase = "1.3"
 DensityInterface = "0.4"
 Distributions = "0.25"
 DynamicIterators = "0.4"
-FLoops = "0.2"
 FillArrays = "0.12, 0.13"
 IfElse = "0.1"
 Infinities = "0.1"
 InverseFunctions = "0.1"
 KeywordCalls = "0.2"
-LazyArrays = "0.21, 0.22"
 LogExpFunctions = "0.3.3"
 MLStyle = "0.4"
 MacroTools = "0.5"
 MappedArrays = "0.4"
-MeasureBase = "0.7"
+MeasureBase = "0.9"
 NamedTupleTools = "0.13, 0.14"
 NestedTuples = "0.3"
 PositiveFactorizations = "0.2"
@@ -68,7 +66,8 @@ Reexport = "1"
 SpecialFunctions = "1, 2"
 Static = "0.5, 0.6"
 StaticArrays = "1.3"
-StatsFuns = "0.9"
+StatsBase = "0.32, 0.33"
+StatsFuns = "0.9, 1"
 TransformVariables = "0.5, 0.6"
 Tricks = "0.1"
 julia = "1.6"

docs/src/affine.md

@@ -69,17 +69,17 @@ Here the `- logdet(σ)` is the "log absolute Jacobian", required to account for
 
 The above requires solving a linear system, which adds some overhead. Even with the convenience of a lower triangular system, it's still not quite as efficient as multiplication.
 
-In addition to the covariance ``Σ``, it's also common to parameterize a multivariate normal by its _precision matrix_, defined as the inverse of the covariance matrix, ``Ω = Σ⁻¹``. Similar to our use of ``σ`` for the lower Cholesky factor of `Σ`, we'll use ``ω`` for the lower Cholesky factor of ``Ω``.
+In addition to the covariance ``Σ``, it's also common to parameterize a multivariate normal by its _precision matrix_, defined as the inverse of the covariance matrix, ``Ω = Σ⁻¹``. Similar to our use of ``σ`` for the lower Cholesky factor of `Σ`, we'll use ``λ`` for the lower Cholesky factor of ``Ω``.
 
 This parameterization enables more efficient calculation of the log-density using only multiplication and addition,
 
 ```julia
-logdensity_def(d::Normal{(:μ,:ω)}, x) = logdensity_def(d.ω * (x - d.μ)) + logdet(d.ω)
+logdensity_def(d::Normal{(:μ,:λ)}, x) = logdensity_def(d.λ * (x - d.μ)) + logdet(d.λ)
 ```
 
 ## `AffineTransform`
 
-Transforms like ``z → σ z + μ`` and ``z → ω \ z + μ`` can be specified in MeasureTheory.jl using an `AffineTransform`. For example,
+Transforms like ``z → σ z + μ`` and ``z → λ \ z + μ`` can be specified in MeasureTheory.jl using an `AffineTransform`. For example,
 
 ```julia
 julia> f = AffineTransform((μ=3.,σ=2.))
@@ -91,11 +91,11 @@ julia> f(1.0)
 
 In the univariate case this is relatively simple to invert. But if `σ` is a matrix, matrix inversion becomes necessary. This is not always possible as lower triangular matrices are not closed under matrix inversion and as such are not guaranteed to exist. 
 
-With multiple parameterizations of a given family of measures, we can work around these issues. The inverse transform of a ``(μ,σ)`` transform will be in terms of ``(μ,ω)``, and vice-versa. So
+With multiple parameterizations of a given family of measures, we can work around these issues. The inverse transform of a ``(μ,σ)`` transform will be in terms of ``(μ,λ)``, and vice-versa. So
 
 ```julia
 julia> f⁻¹ = inverse(f)
-AffineTransform{(:μ, :ω), Tuple{Float64, Float64}}((μ = -1.5, ω = 2.0))
+AffineTransform{(:μ, :λ), Tuple{Float64, Float64}}((μ = -1.5, λ = 2.0))
 
 julia> f(f⁻¹(4))
 4.0

src/MeasureTheory.jl

@@ -1,21 +1,21 @@
 module MeasureTheory
 
 using Random
-using FLoops
 
 using MeasureBase
 using MLStyle
 using NestedTuples
 import TransformVariables
 const TV = TransformVariables
 
-using TransformVariables: as, asℝ₊, as𝕀, asℝ
+using TransformVariables: asℝ₊, as𝕀, asℝ, transform
 
 import Base
 import Distributions
 const Dists = Distributions
 
 export TV
+export transform
 export ≪
 export gentype
 export For
@@ -30,7 +30,7 @@ export CountingMeasure
 export TrivialMeasure
 export Likelihood
 export testvalue
-export basekleisli
+export basekernel
 
 using Infinities
 using DynamicIterators
@@ -45,15 +45,14 @@ import LogExpFunctions
 import NamedTupleTools
 import InverseFunctions: inverse
 export inverse
-export ifelse
 
 import MeasureBase: insupport, instance, marginals
 import MeasureBase:
     testvalue,
     logdensity_def,
     density_def,
     basemeasure,
-    kleisli,
+    kernel,
     params,
     paramnames,
     ∫,
@@ -62,7 +61,13 @@ import MeasureBase:
 import MeasureBase: ≪
 using MeasureBase: BoundedInts, BoundedReals, CountingMeasure, IntegerDomain, IntegerNumbers
 using MeasureBase: weightedmeasure, restrict
-using MeasureBase: AbstractKleisli
+using MeasureBase: AbstractTransitionKernel
+
+import Statistics: mean, var, std
+
+import MeasureBase: likelihood
+export likelihood
+export log_likelihood_ratio
 
 using StaticArrays
 
@@ -74,8 +79,8 @@ import Base: rand
 
 using Reexport
 @reexport using MeasureBase
-import IfElse: ifelse
-@reexport using IfElse
+import IfElse
+using IfElse
 
 using Tricks: static_hasmethod
 
@@ -87,6 +92,7 @@ export AffineTransform
 export insupport
 export For
 
+using MeasureBase: kernel
 using MeasureBase: Returns
 import MeasureBase: proxy, @useproxy
 import MeasureBase: basemeasure_depth
@@ -101,10 +107,6 @@ gentype(μ::AbstractMeasure) = typeof(testvalue(μ))
 
 # gentype(μ::AbstractMeasure) = gentype(basemeasure(μ))
 
-import Distributions: logpdf, pdf
-
-export pdf, logpdf
-
 xlogx(x::Number) = LogExpFunctions.xlogx(x)
 xlogx(x, y) = x * log(x)
 
@@ -114,6 +116,8 @@ xlogy(x, y) = x * log(y)
 xlog1py(x::Number, y::Number) = LogExpFunctions.xlog1py(x, y)
 xlog1py(x, y) = x * log(1 + y)
 
+as(args...; kwargs...) = TV.as(args...; kwargs...)
+
 include("utils.jl")
 include("const.jl")
 include("combinators/for.jl")
@@ -126,7 +130,6 @@ include("combinators/weighted.jl")
 include("combinators/product.jl")
 include("combinators/transforms.jl")
 include("combinators/exponential-families.jl")
-
 include("resettable-rng.jl")
 include("realized.jl")
 include("combinators/chain.jl")
@@ -151,6 +154,9 @@ include("parameterized/binomial.jl")
 include("parameterized/multinomial.jl")
 include("parameterized/lkj-cholesky.jl")
 include("parameterized/negativebinomial.jl")
+include("parameterized/gamma.jl")
+include("parameterized/snedecorf.jl")
+include("parameterized/inverse-gaussian.jl")
 
 include("combinators/ifelse.jl")