Refactor (#178)

* 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* drop outdated Dists.logpdf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

* tests passing

* drop old tbasemeasure_type code

* update logdensity_def(d::Affine{(:σ,)}, x)

* update license

* Move `solve` to MeasureTheory

* better type stability

* `solve!` moved to MeasureTheory

* exponential families

* work on exponential families

* ExpFamMeasure

* applying an exponential family

* kwstruct for Laplace

* update README (license description)

* kwstruct for Laplace

* kwstruct for Bernoulli

* Simplify logdensity_def(::DynamicFor{M}, y) for now

* using ConcreteStructs

* update dependencies

* inner constructors

* inline logjac

* exponential families and tweedie distributions

* `mydot`

* Soss tests passing

* tests passing

* inline

* condiional meaures

* update affine smart constructor

* update for type stability

* tests passing!

* updates

* Make Bernoulli more efficient

* license

* compat

* Run CI on Juila 1.6

* fix typo

* bump version

Author: cscherrer

.JuliaFormatter.toml

@@ -0,0 +1,20 @@
+indent = 4
+margin = 92
+always_for_in = true
+whitespace_typedefs = false
+whitespace_ops_in_indices = false
+remove_extra_newlines = true
+import_to_using = false
+pipe_to_function_call = false
+short_to_long_function_def = false
+always_use_return = false
+whitespace_in_kwargs = true
+annotate_untyped_fields_with_any = false
+format_docstrings = false
+align_struct_field = true
+align_conditional = true
+align_assignment = true
+align_pair_arrow = true
+conditional_to_if = true
+normalize_line_endings = "unix"
+align_matrix = false

.github/workflows/ci.yml

@@ -22,7 +22,7 @@ jobs:
       fail-fast: false
       matrix:
         version:
-          - '1.5'
+          - '1.6'
           - '1'
         os:
           - ubuntu-latest

LICENSE

@@ -18,4 +18,4 @@ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+SOFTWARE.

Project.toml

@@ -1,18 +1,24 @@
 name = "MeasureTheory"
 uuid = "eadaa1a4-d27c-401d-8699-e962e1bbc33b"
 authors = ["Chad Scherrer <[email protected]> and contributors"]
-version = "0.13.2"
+version = "0.14.0"
 
 [deps]
 Accessors = "7d9f7c33-5ae7-4f3b-8dc6-eff91059b697"
+ChangesOfVariables = "9e997f8a-9a97-42d5-a9f1-ce6bfc15e2c0"
+Compat = "34da2185-b29b-5c13-b0c7-acf172513d20"
 ConcreteStructs = "2569d6c7-a4a2-43d3-a901-331e8e4be471"
 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"
-InfiniteArrays = "4858937d-0d70-526a-a4dd-2d5cb5dd786c"
+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"
@@ -26,32 +32,42 @@ PrettyPrinting = "54e16d92-306c-5ea0-a30b-337be88ac337"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b"
+Static = "aedffcd0-7271-4cad-89d0-dc628f76c6d3"
+StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
 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"
 ConcreteStructs = "0.2"
 ConstructionBase = "1.3"
+DensityInterface = "0.4"
 Distributions = "0.25"
 DynamicIterators = "0.4"
-FillArrays = "0.12"
-InfiniteArrays = "0.11, 0.12"
+FLoops = "0.2"
+FillArrays = "0.12, 0.13"
+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 = "0.5"
-NamedTupleTools = "0.13"
+MeasureBase = "0.6"
+NamedTupleTools = "0.13, 0.14"
 NestedTuples = "0.3"
 PositiveFactorizations = "0.2"
-PrettyPrinting = "0.3"
+PrettyPrinting = "0.3, 0.4"
 Reexport = "1"
-SpecialFunctions = "1"
+SpecialFunctions = "1, 2"
+Static = "0.5, 0.6"
+StaticArrays = "1.3"
 StatsFuns = "0.9"
-TransformVariables = "0.4"
+TransformVariables = "0.5, 0.6"
 Tricks = "0.1"
 julia = "1.5"
 

docs/src/adding.md

@@ -33,11 +33,11 @@ The `≪` (typed as `\ll <TAB>`) can be read "is dominated by". This just means
 
 ### Defining a Log Density
 
-Most computations involve log-densities rather than densities themselves, so these are our first priority. `density(d,x)` will default to `exp(logdensity(d,x))`, but you can add a separate method if it's more efficient.
+Most computations involve log-densities rather than densities themselves, so these are our first priority. `density(d,x)` will default to `exp(logdensity_def(d,x))`, but you can add a separate method if it's more efficient.
 
 The definition is simple:
 ```julia
-logdensity(d::Normal{()} , x) = - x^2 / 2 
+logdensity_def(d::Normal{()} , x) = - x^2 / 2 
 ```
 
 There are a few things here worth noting.
@@ -53,7 +53,7 @@ Also, there's nothing here about `μ` and `σ`. These _location-scale parameters
 Let's look at another example, the Beta distribution. Here the base measure is `Lebesgue(𝕀)` (support is the unit interval). The log-density is
 
 ```julia
-function logdensity(d::Beta{(:α, :β)}, x)
+@inline function logdensity_def(d::Beta{(:α, :β)}, x)
     return (d.α - 1) * log(x) + (d.β - 1) * log(1 - x) - logbeta(d.α, d.β)
 end
 ```
@@ -114,15 +114,15 @@ basemeasure(μ::Lebesgue) = μ
 
 isprimitive(::Lebesgue) = true
 
-sampletype(::Lebesgue{ℝ}) = Float64
-sampletype(::Lebesgue{ℝ₊}) = Float64
-sampletype(::Lebesgue{𝕀}) = Float64
+gentype(::Lebesgue{ℝ}) = Float64
+gentype(::Lebesgue{ℝ₊}) = Float64
+gentype(::Lebesgue{𝕀}) = Float64
 
-logdensity(::Lebesgue, x) = zero(float(x))
+logdensity_def(::Lebesgue, x) = zero(float(x))
 ```
 
-We haven't yet talked about `sampletype`. When you call `rand` without specifying a type, there needs to be a default. That default is the `sampletype`. This only needs to be defined for primitive measures, because others will fall back on 
+We haven't yet talked about `gentype`. When you call `rand` without specifying a type, there needs to be a default. That default is the `gentype`. This only needs to be defined for primitive measures, because others will fall back on 
 
 ```julia
-sampletype(μ::AbstractMeasure) = sampletype(basemeasure(μ))
+gentype(μ::AbstractMeasure) = gentype(basemeasure(μ))
 ```

docs/src/affine.md

@@ -62,7 +62,7 @@ z = σ⁻¹ (x - μ)
 so the log-density is
 
 ```julia
-logdensity(d::Normal{(:μ,:σ)}, x) = logdensity(d.σ \ (x - d.μ)) - logdet(d.σ)
+logdensity_def(d::Normal{(:μ,:σ)}, x) = logdensity_def(d.σ \ (x - d.μ)) - logdet(d.σ)
 ```
 
 Here the `- logdet(σ)` is the "log absolute Jacobian", required to account for the stretching of the space.
@@ -74,7 +74,7 @@ In addition to the covariance ``Σ``, it's also common to parameterize a multiva
 This parameterization enables more efficient calculation of the log-density using only multiplication and addition,
 
 ```julia
-logdensity(d::Normal{(:μ,:ω)}, x) = logdensity(d.ω * (x - d.μ)) + logdet(d.ω)
+logdensity_def(d::Normal{(:μ,:ω)}, x) = logdensity_def(d.ω * (x - d.μ)) + logdet(d.ω)
 ```
 
 ## `AffineTransform`
@@ -94,7 +94,7 @@ In the univariate case this is relatively simple to invert. But if `σ` is a mat
 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
 
 ```julia
-julia> f⁻¹ = inv(f)
+julia> f⁻¹ = inverse(f)
 AffineTransform{(:μ, :ω), Tuple{Float64, Float64}}((μ = -1.5, ω = 2.0))
 
 julia> f(f⁻¹(4))

docs/src/old_readme.md

@@ -153,21 +153,21 @@ A density doesn't exist by itself, but is defined relative to some _base measure
 ```julia
 @implement Density{Normal{X,P},X} where {X, P <: NamedTuple{(:μ, :σ)}} begin
     basemeasure(d) = Lebesgue(X)
-    logdensity(d, x) = - (log(2) + log(π)) / 2 - log(d.par.σ)  - (x - d.par.μ)^2 / (2 * d.par.σ^2)
+    logdensity_def(d, x) = - (log(2) + log(π)) / 2 - log(d.par.σ)  - (x - d.par.μ)^2 / (2 * d.par.σ^2)
 end
 ```
 
 Now we can compute the log-density:
 
 ```julia
-julia> logdensity(Normal(0.0, 0.5), 1.0)
+julia> logdensity_def(Normal(0.0, 0.5), 1.0)
 -2.2257913526447273
 ```
 
 And just to check that our default is working,
 
 ```julia
-julia> logdensity(Normal(μ=0.0, σ=0.5), 1.0)
+julia> logdensity_def(Normal(μ=0.0, σ=0.5), 1.0)
 -2.2257913526447273
 ```
 
@@ -178,14 +178,14 @@ eltype(::Type{Normal}, ::Type{NamedTuple{(:μ, :τ), Tuple{A, B}}}) where {A,B}
 
 @implement Density{Normal{X,P},X} where {X, P <: NamedTuple{(:μ, :τ)}} begin
     basemeasure(d) = Lebesgue(X)
-    logdensity(d, x) = - (log(2) + log(π) - log(d.par.τ)  + d.par.τ * (x - d.par.μ)^2) / 2
+    logdensity_def(d, x) = - (log(2) + log(π) - log(d.par.τ)  + d.par.τ * (x - d.par.μ)^2) / 2
 end
 ```
 
 And another check:
 
 ```julia
-julia> logdensity(Normal(μ=0.0, τ=4.0), 1.0)
+julia> logdensity_def(Normal(μ=0.0, τ=4.0), 1.0)
 -2.2257913526447273
 ```
 

jet-test.jl

@@ -0,0 +1,5 @@
+using MeasureTheory
+
+dist = For(3) do j Normal(σ=j) end
+
+b = basemeasure_depth(dist)

scratch/orderstats.jl

@@ -2,11 +2,11 @@ struct OrderStatistic{n,r,B} <: AbstractMeasure
     base::B
 end
 
-function basemeasure(d::OrderStatistic{n,r})
+@inline function basemeasure(d::OrderStatistic{n,r})
     WeightedMeasure(logbeta(n+1, r+1), d.base)
 end
 
-function logdensity(d::OrderStatistic{n,r}, x)
+@inline function logdensity_def(d::OrderStatistic{n,r}, x)
     logF(x) = logcdf(d.base, x)
     logFᶜ(x) = logccdf(d.base, x)
     return (r - 1) * logF(x) + (n - r) * logFᶜ(x)