Add Unicode alias \oplus for convolve

Project.toml

@@ -1,7 +1,7 @@
 name = "Distributions"
 uuid = "31c24e10-a181-5473-b8eb-7969acd0382f"
 authors = ["JuliaStats"]
-version = "0.25.34"
+version = "0.25.35"
 
 [deps]
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"

docs/make.jl

@@ -16,6 +16,7 @@ makedocs(
         "reshape.md",
         "cholesky.md",
         "mixture.md",
+        "convolution.md",
         "fit.md",
         "extends.md",
         "density_interface.md",

docs/src/convolution.md

@@ -0,0 +1,12 @@
+# Convolutions
+
+A [convolution of two probability distributions](https://en.wikipedia.org/wiki/List_of_convolutions_of_probability_distributions)
+is the probability distribution of the sum of two independent random variables that are
+distributed according to these distributions.
+
+The convolution of two distributions can be constructed with [`convolve`](@ref) or its
+Unicode alias [`⊕`](@ref).
+
+```@docs
+convolve
+```

src/Distributions.jl

@@ -190,6 +190,7 @@ export
     componentwise_logpdf,   # component-wise logpdf for mixture models
     concentration,      # the concentration parameter
     convolve,           # convolve distributions of the same type
+    ⊕,                  # unicode alias of convolve
     dim,                # sample dimension of multivariate distribution
     dof,                # get the degree of freedom
     entropy,            # entropy of distribution in nats

src/convolution.jl

@@ -1,25 +1,33 @@
 """
-    convolve(d1::T, d2::T) where T<:Distribution -> Distribution
-
-Convolve two distributions of the same type to yield the distribution corresponding to the
-sum of independent random variables drawn from the underlying distributions.
-
-The function is only defined in the cases where the convolution has a closed form as
-defined here https://en.wikipedia.org/wiki/List_of_convolutions_of_probability_distributions
-
-* `Bernoulli`
-* `Binomial`
-* `NegativeBinomial`
-* `Geometric`
-* `Poisson`
-* `Normal`
-* `Cauchy`
-* `Chisq`
-* `Exponential`
-* `Gamma`
-* `MultivariateNormal`
+    convolve(d1::Distribution, d2::Distribution)
+    d1 ⊕ d2
+
+Convolve two distributions and return the distribution corresponding to the sum of
+independent random variables drawn from the underlying distributions.
+
+The Unicode operator `⊕` can be typed by `\\oplus<tab>`.
+
+Currently, the function is only defined in cases where the convolution has a closed form.
+More precisely, the function is defined if the distributions of `d1` and `d2` are the same
+and one of
+* [`Bernoulli`](@ref)
+* [`Binomial`](@ref)
+* [`NegativeBinomial`](@ref)
+* [`Geometric`](@ref)
+* [`Poisson`](@ref)
+* [`Normal`](@ref)
+* [`Cauchy`](@ref)
+* [`Chisq`](@ref)
+* [`Exponential`](@ref)
+* [`Gamma`](@ref)
+* [`MvNormal`](@ref)
+
+External links: [List of convolutions of probability distributions on Wikipedia](https://en.wikipedia.org/wiki/List_of_convolutions_of_probability_distributions)
 """
-function convolve end
+convolve(::Distribution, ::Distribution)
+
+# define Unicode alias
+const ⊕ = convolve
 
 # discrete univariate
 function convolve(d1::Bernoulli, d2::Bernoulli)
@@ -61,27 +69,9 @@ function convolve(d1::Gamma, d2::Gamma)
 end
 
 # continuous multivariate
-# The first two methods exist for performance reasons to avoid unnecessarily converting
-# PDMats to a Matrix
-function convolve(
-    d1::Union{IsoNormal, ZeroMeanIsoNormal, DiagNormal, ZeroMeanDiagNormal},
-    d2::Union{IsoNormal, ZeroMeanIsoNormal, DiagNormal, ZeroMeanDiagNormal},
-    )
-    _check_convolution_shape(d1, d2)
-    return MvNormal(d1.μ .+ d2.μ, d1.Σ + d2.Σ)
-end
-
-function convolve(
-    d1::Union{FullNormal, ZeroMeanFullNormal},
-    d2::Union{FullNormal, ZeroMeanFullNormal},
-    )
-    _check_convolution_shape(d1, d2)
-    return MvNormal(d1.μ .+ d2.μ, d1.Σ.mat + d2.Σ.mat)
-end
-
 function convolve(d1::MvNormal, d2::MvNormal)
     _check_convolution_shape(d1, d2)
-    return MvNormal(d1.μ .+ d2.μ, Matrix(d1.Σ) + Matrix(d2.Σ))
+    return MvNormal(d1.μ + d2.μ, d1.Σ + d2.Σ)
 end