Switch to FFTA

This change adds support for more element types and drops a binary
GPL dependency. For small problems, there is a minor slowdown, but
it goes away for slightly larger problems where the processing between
the ffts start dominating.

Author: andreasnoack

Project.toml

@@ -1,19 +1,22 @@
 name = "KernelDensity"
 uuid = "5ab0869b-81aa-558d-bb23-cbf5423bbe9b"
-authors = ["Simon Byrne and various contributors"]
 version = "0.6.10"
+authors = ["Simon Byrne and various contributors"]
 
 [deps]
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 DocStringExtensions = "ffbed154-4ef7-542d-bbb7-c09d3a79fcae"
-FFTW = "7a1cc6ca-52ef-59f5-83cd-3a7055c09341"
+FFTA = "b86e33f2-c0db-4aa1-a6e0-ab43e668529e"
 Interpolations = "a98d9a8b-a2ab-59e6-89dd-64a1c18fca59"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 
+[sources]
+FFTA = {rev = "an/rfftas2cfft", url = "https://github.com/JuliaMath/FFTA.jl.git"}
+
 [compat]
 Distributions = "0.23, 0.24, 0.25"
 DocStringExtensions = "0.8, 0.9"
-FFTW = "1"
+FFTA = "0.3"
 Interpolations = "0.9, 0.10, 0.11, 0.12, 0.13, 0.14, 0.15, 0.16"
 StatsBase = "0.33, 0.34"
 julia = "1"

src/KernelDensity.jl

@@ -6,7 +6,7 @@ using Distributions
 using Interpolations
 
 import Distributions: twoπ, pdf
-import FFTW: rfft, irfft
+import FFTA: rfft, irfft
 
 export kde, kde_lscv, UnivariateKDE, BivariateKDE, InterpKDE, pdf
 

src/bivariate.jl

@@ -79,7 +79,7 @@ end
 function conv(k::BivariateKDE, dist::Tuple{UnivariateDistribution,UnivariateDistribution})
     # Transform to Fourier basis
     Kx, Ky = size(k.density)
-    ft = rfft(k.density)
+    ft::Matrix{Complex{Float64}} = rfft(k.density)
 
     distx, disty = dist
 

src/univariate.jl

@@ -123,7 +123,7 @@ end
 function conv(k::UnivariateKDE, dist::UnivariateDistribution)
     # Transform to Fourier basis
     K = length(k.density)
-    ft = rfft(k.density)
+    ft::Vector{Complex{Float64}} = rfft(k.density)
 
     # Convolve fft with characteristic function of kernel
     # empirical cf
@@ -152,7 +152,7 @@ end
 
 	kde(data; kwargs...)
 	kde((xdata, ydata); kwargs...)
-	
+
 Kernel density estimation method. Returns 1D or 2D KDE object. The grid used and the values of the estimated density can be obtained from fields `.x` and `.density` respectively. To obtain kde values at points different than the initial grid use the `pdf` method.
 
 The keyword arguments are

test/bivariate.jl

@@ -4,7 +4,7 @@ using KernelDensity
 
 import KernelDensity: kernel_dist, default_bandwidth, kde_boundary, kde_range, tabulate
 
-for D in [Tuple{Normal,Normal}, Tuple{Uniform,Uniform}, Tuple{Logistic,Logistic},
+@testset "kernel_dist: D=$D" for D in [Tuple{Normal,Normal}, Tuple{Uniform,Uniform}, Tuple{Logistic,Logistic},
           (Normal, Normal), (Uniform, Uniform), (Logistic, Logistic)]
     d = KernelDensity.kernel_dist(D,(0.5,0.5))
     dx,dy = d
@@ -17,7 +17,7 @@ end
 r = kde_range((-2.0,2.0), 128)
 @test step(r) > 0
 
-for X in ([0.0], [0.0,0.0], [0.0,0.5], [-0.5:0.1:0.5;])
+@testset "X=$X" for X in ([0.0], [0.0,0.0], [0.0,0.5], [-0.5:0.1:0.5;])
     w = default_bandwidth(X)
     @test w > 0
     lo, hi = kde_boundary(X,w)
@@ -62,8 +62,10 @@ for X in ([0.0], [0.0,0.0], [0.0,0.5], [-0.5:0.1:0.5;])
 end
 
 k11 = kde([0.0 0.0; 1.0 1.0], (r,r), bandwidth=(1,1), weights=[0,1])
-k12 = kde([1.0 1.0], (r,r), bandwidth=(1,1))
-@test k11.density ≈ k12.density
+@testset "weight argument" begin
+    k12 = kde([1.0 1.0], (r,r), bandwidth=(1,1))
+    @test k11.density ≈ k12.density
+end
 
 @testset "broadcasting. Issue 63" begin
     xyvals = [0.0, 1.0]

test/runtests.jl

@@ -1,13 +1,12 @@
+using Test
+
 tests = [
          "univariate",
          "bivariate",
          "interp",
          ]
 
-println("Running tests:")
-
-for t in tests
+@testset "$t" for t in tests
     test_fn = "$t.jl"
-    println(" * $test_fn")
     include(test_fn)
 end