kmedoids(): fix duplicate medoids handling

If medoids have duplicate coordinates, make
sure the medoid is assigned to itself,
otherwise its cluster might be empty.
If the distance to itself is further than to
another medoid, it is an issue with the distance
matrix, and ArgumentError is raised.

fixes #231

Author: alyst

src/kmedoids.jl

@@ -206,7 +206,8 @@ function _kmed_update_assignments!(dist::AbstractMatrix{<:Real}, # in: (n, n)
             m = medoids[i]
             v = dist[m, j]
             # assign if current medoid is closer or if it is j itself
-            if v < mv
+            if (v < mv) || (m == j)
+                (v <= mv) || throw(ArgumentError("sample #$j reassigned from medoid[$p]=#$(medoids[p]) (distance=$mv) to medoid[$i]=#$m (distance=$v); check the distance matrix correctness"))
                 p = i
                 mv = v
             end

test/kmedoids.jl

@@ -9,8 +9,12 @@ include("test_helpers.jl")
     Random.seed!(34568)
     @test_throws ArgumentError kmedoids(randn(2, 3), 1)
     @test_throws ArgumentError kmedoids(randn(2, 3), 4)
-    dist = inv.(max.(pairwise(Euclidean(), randn(2, 3), dims=2), 0.1))
+    dist = max.(pairwise(Euclidean(), randn(2, 3), dims=2), 0.1)
     @test @inferred(kmedoids(dist, 2)) isa KmedoidsResult
+    # incorrect distance matrix
+    invdist = inv.(max.(pairwise(Euclidean(), randn(2, 3), dims=2), 0.1))
+    @test_throws ArgumentError kmedoids(invdist, 2)
+
     @test_throws ArgumentError kmedoids(dist, 2, display=:mylog)
     for disp in keys(Clustering.DisplayLevels)
         @test @inferred(kmedoids(dist, 2, display=disp)) isa KmedoidsResult
@@ -48,6 +52,15 @@ R = @inferred(kmedoids(dist, k))
     end
 end
 
+@testset "Duplicated points (#231)" begin
+    pts = [0.0 0.0]
+    dists = pairwise(SqEuclidean(), pts, dims=2)
+    dupmed = kmedoids(dists, 2)
+    @test nclusters(dupmed) == 2
+    @test sort(dupmed.medoids) == [1, 2]
+    @test sort(dupmed.assignments) == [1, 2]
+end
+
 @testset "Toy example #1" begin
     pts = [1 2 3; .1 .2 .3; 4 5.6 7]
     # k=1 and k=n cases