support addition & subtraction

é

src/vector.jl

@@ -82,3 +82,22 @@ end
 
 SparseArrays.nnz(v::DynamicSparseVector) = nnz(v.pma)
 Base.copy(::DynamicSparseVector) = error("copy of a dynamic sparse vector not implemented.")
+
+# to use linear algebra:
+function SparseArrays.nonzeroinds(v::DynamicSparseVector{K,T}) where {K,T}
+    return reduce(v.pma.array; init = K[]) do collection, e
+        if !isnothing(e)
+            push!(collection, first(e))
+        end
+        return collection
+    end
+end
+
+function SparseArrays.nonzeros(v::DynamicSparseVector{K,T}) where {K,T}
+    return reduce(v.pma.array; init = T[]) do collection, e
+        if !isnothing(e)
+            push!(collection, last(e))
+        end
+        return collection
+    end
+end 

test/functional/functionaltests.jl

@@ -38,8 +38,10 @@ function functional_tests()
         dynsparsematrix_fill_mode()
     end
 
-    @testset "spMv" begin
+    @testset "lin alg" begin
         spMv1()
+        addition()
+        subtraction()
     end
     return
 end

test/functional/math.jl

@@ -50,3 +50,45 @@ function spMv1()
     @test i == j == d
 end
 
+function addition()
+    vec_rows1 = rand(rng, 1:100, 25)
+    vec_values1 = rand(rng, 1:10, 25)
+
+    dyn_vec1 = dynamicsparsevec(vec_rows1, vec_values1, 100)
+    vec1 = sparsevec(vec_rows1, vec_values1, 100)
+
+    vec_rows2 = rand(rng, 1:100, 25)
+    vec_values2 = rand(rng, 1:10, 25)
+
+    dyn_vec2 = dynamicsparsevec(vec_rows2, vec_values2, 100)
+    vec2 = sparsevec(vec_rows2, vec_values2, 100)
+
+    classic_sum = vec1 + vec2
+
+    @test dyn_vec1 + dyn_vec2 == classic_sum
+    @test vec1 + dyn_vec2 == classic_sum
+    @test dyn_vec1 + vec2 == classic_sum
+end
+
+function subtraction()
+    vec_rows1 = rand(rng, 1:100, 25)
+    vec_values1 = rand(rng, 1:10, 25)
+
+    dyn_vec1 = dynamicsparsevec(vec_rows1, vec_values1, 100)
+    vec1 = sparsevec(vec_rows1, vec_values1, 100)
+
+    vec_rows2 = rand(rng, 1:100, 25)
+    vec_values2 = rand(rng, 1:10, 25)
+
+    dyn_vec2 = dynamicsparsevec(vec_rows2, vec_values2, 100)
+    vec2 = sparsevec(vec_rows2, vec_values2, 100)
+
+    classic_sub = vec1 - vec2
+
+    @test dyn_vec1 - dyn_vec2 == classic_sub
+    @test vec1 - dyn_vec2 == classic_sub
+    @test dyn_vec1 - vec2 == classic_sub
+    @test -dyn_vec1 == -vec1
+    @test -dyn_vec2 == -vec2
+end
+