adds farthest first traversal fft

src/SimilaritySearch.jl

@@ -94,6 +94,7 @@ include("deprecated.jl")
 
 include("allknn.jl")
 include("neardup.jl")
+include("fft.jl")
 include("closestpair.jl")
 include("hsp.jl")
 

src/fft.jl

@@ -0,0 +1,49 @@
+# This file is a part of SimilaritySearch.jl
+
+export fft
+
+"""
+    fft(dist::SemiMetric, X::AbstractDatabase, k; verbose=true)
+
+Selects `k` items far from each other based on Farthest First Traversal algorithm.
+
+Returns a named tuple with the following fields:
+- `centers` contains the list of centers (indexes to ``X``)
+- `nn` the id of the nearest center (in ``X`` order, identifiers between 1 to `length(X))
+- `nndists` the distance from each object in the database to its nearest centers (in ``X`` order)
+- `dmax` smallest distance among centers
+
+Based on `enet.jl` from `KCenters.jl`
+"""
+function fft(dist::SemiMetric, X::AbstractDatabase, k::Integer; verbose=true)
+    N = length(X)
+    centers = Int32[]
+    dmaxlist = Float32[]
+    nndists = Vector{Float32}(undef, N)
+    fill!(nndists, typemax(Float32))
+    nn = zeros(UInt32, N) 
+    imax::Int = rand(1:N)
+    dmax::Float32 = typemax(Float32)
+    N == 0 && return (; centers, nn, dists=nndists, dmax)
+
+    @inbounds for i in 1:N
+        push!(dmaxlist, dmax)
+        push!(centers, imax)
+        verbose && println(stderr, "computing fartest point $(length(centers)), dmax: $dmax, imax: $imax, n: $(length(X))")
+
+        pivot = X[imax]
+        @batch minbatch=getminbatch(0, N) for i in 1:N
+            d = evaluate(dist, X[i], pivot)
+            if d < nndists[i]
+                nndists[i] = d
+                nn[i] = imax
+            end
+        end
+
+        dmax, imax = findmax(nndists)
+        length(dmaxlist) < k || break
+    end
+
+    (; centers, nn, dists=nndists, dmax)
+end
+

test/runtests.jl

@@ -23,5 +23,6 @@ include("testadj.jl")
 include("testallknn.jl")
 include("testhsp.jl")
 include("testneardup.jl")
+include("testfft.jl")
 include("testclosestpair.jl")
 include("testsearchgraph.jl")

test/testfft.jl

@@ -0,0 +1,12 @@
+# This file is a part of SimilaritySearch.jl
+
+using Test, SimilaritySearch, LinearAlgebra
+
+@testset "farthest first traversal" begin 
+    dist = L2Distance()
+    X = rand(Float32, 4, 300)
+    res = fft(dist, MatrixDatabase(X), 30)
+    @test Set(res.centers) == Set(res.nn)
+    @test all(res.dmax .>= res.dists)
+end
+