Add GraphsSharedArraysExt and remove Distributed dependency

Project.toml

@@ -6,7 +6,6 @@ version = "1.12.1"
 ArnoldiMethod = "ec485272-7323-5ecc-a04f-4719b315124d"
 Compat = "34da2185-b29b-5c13-b0c7-acf172513d20"
 DataStructures = "864edb3b-99cc-5e75-8d2d-829cb0a9cfe8"
-Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
 Inflate = "d25df0c9-e2be-5dd7-82c8-3ad0b3e990b9"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
@@ -15,6 +14,12 @@ SimpleTraits = "699a6c99-e7fa-54fc-8d76-47d257e15c1d"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 
+[weakdeps]
+SharedArrays = "1a1011a3-84de-559e-8e89-a11a2f7dc383"
+
+[extensions]
+GraphsSharedArraysExt = "SharedArrays"
+
 [compat]
 Aqua = "0.6"
 ArnoldiMethod = "0.4"
@@ -32,17 +37,19 @@ julia = "1.6"
 Aqua = "4c88cf16-eb10-579e-8560-4a9242c79595"
 Base64 = "2a0f44e3-6c83-55bd-87e4-b1978d98bd5f"
 DelimitedFiles = "8bb1440f-4735-579b-a4ab-409b98df4dab"
+Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
 Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
 JET = "c3a54625-cd67-489e-a8e7-0a5a0ff4e31b"
 JuliaFormatter = "98e50ef6-434e-11e9-1051-2b60c6c9e899"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 Pkg = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
+SharedArrays = "1a1011a3-84de-559e-8e89-a11a2f7dc383"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 Unitful = "1986cc42-f94f-5a68-af5c-568840ba703d"
 
 [targets]
-test = ["Aqua", "Base64", "DelimitedFiles", "Documenter", "JET", "JuliaFormatter", "LinearAlgebra", "Pkg", "Random", "SparseArrays", "StableRNGs", "Statistics", "Test", "Unitful"]
+test = ["Aqua", "Base64", "DelimitedFiles", "Distributed", "Documenter", "JET", "JuliaFormatter", "LinearAlgebra", "Pkg", "Random", "SharedArrays", "SparseArrays", "StableRNGs", "Statistics", "Test", "Unitful"]

ext/GraphsSharedArraysExt.jl

@@ -0,0 +1,170 @@
+module GraphsSharedArraysExt
+
+using Graphs
+using SharedArrays: SharedArrays, SharedMatrix, SharedVector, sdata
+using SharedArrays.Distributed: @distributed
+using Random: shuffle
+
+# betweenness
+function Graphs.Parallel.distr_betweenness_centrality(
+    g::AbstractGraph,
+    vs=vertices(g),
+    distmx::AbstractMatrix=weights(g);
+    normalize=true,
+    endpoints=false,
+)::Vector{Float64}
+    n_v = nv(g)
+    k = length(vs)
+    isdir = is_directed(g)
+
+    # Parallel reduction
+
+    betweenness = @distributed (+) for s in vs
+        temp_betweenness = zeros(n_v)
+        if degree(g, s) > 0  # this might be 1?
+            state = Graphs.dijkstra_shortest_paths(
+                g, s, distmx; allpaths=true, trackvertices=true
+            )
+            if endpoints
+                Graphs._accumulate_endpoints!(temp_betweenness, state, g, s)
+            else
+                Graphs._accumulate_basic!(temp_betweenness, state, g, s)
+            end
+        end
+        temp_betweenness
+    end
+
+    Graphs._rescale!(betweenness, n_v, normalize, isdir, k)
+
+    return betweenness
+end
+
+# closeness
+function Graphs.Parallel.distr_closeness_centrality(
+    g::AbstractGraph, distmx::AbstractMatrix=weights(g); normalize=true
+)::Vector{Float64}
+    n_v = Int(nv(g))
+    closeness = SharedVector{Float64}(n_v)
+    fill!(closeness, 0.0)
+
+    @sync @distributed for u in vertices(g)
+        if degree(g, u) == 0     # no need to do Dijkstra here
+            closeness[u] = 0.0
+        else
+            d = Graphs.dijkstra_shortest_paths(g, u, distmx).dists
+            δ = filter(x -> x != typemax(x), d)
+            σ = sum(δ)
+            l = length(δ) - 1
+            if σ > 0
+                closeness[u] = l / σ
+                if normalize
+                    n = l * 1.0 / (n_v - 1)
+                    closeness[u] *= n
+                end
+            else
+                closeness[u] = 0.0
+            end
+        end
+    end
+    return sdata(closeness)
+end
+
+# radiality
+function Graphs.Parallel.distr_radiality_centrality(g::AbstractGraph)::Vector{Float64}
+    n_v = nv(g)
+    vs = vertices(g)
+    n = ne(g)
+    meandists = SharedVector{Float64}(Int(n_v))
+    maxdists = SharedVector{Float64}(Int(n_v))
+
+    @sync @distributed for i in 1:n_v
+        d = Graphs.dijkstra_shortest_paths(g, vs[i])
+        maxdists[i] = maximum(d.dists)
+        meandists[i] = sum(d.dists) / (n_v - 1)
+        nothing
+    end
+    dmtr = maximum(maxdists)
+    radialities = collect(meandists)
+    return ((dmtr + 1) .- radialities) ./ dmtr
+end
+
+# stress
+function Graphs.Parallel.distr_stress_centrality(
+    g::AbstractGraph, vs=vertices(g)
+)::Vector{Int64}
+    n_v = nv(g)
+    k = length(vs)
+    isdir = is_directed(g)
+
+    # Parallel reduction
+    stress = @distributed (+) for s in vs
+        temp_stress = zeros(Int64, n_v)
+        if degree(g, s) > 0  # this might be 1?
+            state = Graphs.dijkstra_shortest_paths(g, s; allpaths=true, trackvertices=true)
+            Graphs._stress_accumulate_basic!(temp_stress, state, g, s)
+        end
+        temp_stress
+    end
+    return stress
+end
+
+# generate_reduce
+function Graphs.Parallel.distr_generate_reduce(
+    g::AbstractGraph{T}, gen_func::Function, comp::Comp, reps::Integer
+) where {T<:Integer,Comp}
+    # Type assert required for type stability
+    min_set::Vector{T} = @distributed ((x, y) -> comp(x, y) ? x : y) for _ in 1:reps
+        gen_func(g)
+    end
+    return min_set
+end
+
+# eccentricity
+function Graphs.Parallel.distr_eccentricity(
+    g::AbstractGraph, vs=vertices(g), distmx::AbstractMatrix{T}=weights(g)
+) where {T<:Number}
+    vlen = length(vs)
+    eccs = SharedVector{T}(vlen)
+    @sync @distributed for i in 1:vlen
+        d′ = Graphs.dijkstra_shortest_paths(g, vs[i], distmx)
+        eccs[i] = maximum(d′.dists)
+    end
+    d = sdata(eccs)
+    maximum(d) == typemax(T) && @warn("Infinite path length detected")
+    return d
+end
+
+# dijkstra shortest paths
+function Graphs.Parallel.distr_dijkstra_shortest_paths(
+    g::AbstractGraph{U}, sources=vertices(g), distmx::AbstractMatrix{T}=weights(g)
+) where {T<:Number} where {U}
+    n_v = nv(g)
+    r_v = length(sources)
+
+    # TODO: remove `Int` once julialang/#23029 / #23032 are resolved
+    dists = SharedMatrix{T}(Int(r_v), Int(n_v))
+    parents = SharedMatrix{U}(Int(r_v), Int(n_v))
+
+    @sync @distributed for i in 1:r_v
+        state = Graphs.dijkstra_shortest_paths(g, sources[i], distmx)
+        dists[i, :] = state.dists
+        parents[i, :] = state.parents
+    end
+
+    result = Graphs.Parallel.MultipleDijkstraState(sdata(dists), sdata(parents))
+    return result
+end
+
+# random greedy color
+function Graphs.Parallel.distr_random_greedy_color(
+    g::AbstractGraph{T}, reps::Integer
+) where {T<:Integer}
+    best = @distributed (Graphs.best_color) for i in 1:reps
+        seq = shuffle(vertices(g))
+        Graphs.perm_greedy_color(g, seq)
+    end
+
+    return convert(Graphs.Coloring{T}, best)
+end
+
+end

src/Graphs.jl

@@ -560,5 +560,9 @@ include("Experimental/Experimental.jl")
 include("Parallel/Parallel.jl")
 include("Test/Test.jl")
 
+if !isdefined(Base, :get_extension)
+    include("../ext/GraphsSharedArraysExt.jl")
+end
+
 using .LinAlg
 end # module

src/Parallel/Parallel.jl

@@ -2,9 +2,7 @@ module Parallel
 
 using Graphs
 using Graphs: sample, AbstractPathState, JohnsonState, BellmanFordState, FloydWarshallState
-using Distributed: @distributed
 using Base.Threads: @threads, nthreads, Atomic, atomic_add!, atomic_cas!
-using SharedArrays: SharedMatrix, SharedVector, sdata
 using ArnoldiMethod: LM, SR, LR, partialschur, partialeigen
 using Random: AbstractRNG, shuffle
 import SparseArrays: sparse

src/Parallel/centrality/betweenness.jl

@@ -4,7 +4,7 @@
     distmx::AbstractMatrix=weights(g);
     normalize=true,
     endpoints=false,
-    parallel=:distributed,
+    parallel=:threads,
 )
     return if parallel == :distributed
         distr_betweenness_centrality(
@@ -23,7 +23,7 @@
     distmx::AbstractMatrix=weights(g);
     normalize=true,
     endpoints=false,
-    parallel=:distributed,
+    parallel=:threads,
     rng::Union{Nothing,AbstractRNG}=nothing,
     seed::Union{Nothing,Integer}=nothing,
 )
@@ -39,37 +39,10 @@
     end
 end
 
-function distr_betweenness_centrality(
-    g::AbstractGraph,
-    vs=vertices(g),
-    distmx::AbstractMatrix=weights(g);
-    normalize=true,
-    endpoints=false,
-)::Vector{Float64}
-    n_v = nv(g)
-    k = length(vs)
-    isdir = is_directed(g)
-
-    # Parallel reduction
-
-    betweenness = @distributed (+) for s in vs
-        temp_betweenness = zeros(n_v)
-        if degree(g, s) > 0  # this might be 1?
-            state = Graphs.dijkstra_shortest_paths(
-                g, s, distmx; allpaths=true, trackvertices=true
-            )
-            if endpoints
-                Graphs._accumulate_endpoints!(temp_betweenness, state, g, s)
-            else
-                Graphs._accumulate_basic!(temp_betweenness, state, g, s)
-            end
-        end
-        temp_betweenness
-    end
-
-    Graphs._rescale!(betweenness, n_v, normalize, isdir, k)
-
-    return betweenness
+function distr_betweenness_centrality(args...; kwargs...)
+    return error(
+        "`parallel = :distributed` requested, but SharedArrays or Distributed is not loaded"
+    )
 end
 
 function distr_betweenness_centrality(

src/Parallel/centrality/closeness.jl

@@ -1,8 +1,5 @@
 function closeness_centrality(
-    g::AbstractGraph,
-    distmx::AbstractMatrix=weights(g);
-    normalize=true,
-    parallel=:distributed,
+    g::AbstractGraph, distmx::AbstractMatrix=weights(g); normalize=true, parallel=:threads
 )
     return if parallel == :distributed
         distr_closeness_centrality(g, distmx; normalize=normalize)
@@ -11,33 +8,10 @@
     end
 end
 
-function distr_closeness_centrality(
-    g::AbstractGraph, distmx::AbstractMatrix=weights(g); normalize=true
-)::Vector{Float64}
-    n_v = Int(nv(g))
-    closeness = SharedVector{Float64}(n_v)
-    fill!(closeness, 0.0)
-
-    @sync @distributed for u in vertices(g)
-        if degree(g, u) == 0     # no need to do Dijkstra here
-            closeness[u] = 0.0
-        else
-            d = Graphs.dijkstra_shortest_paths(g, u, distmx).dists
-            δ = filter(x -> x != typemax(x), d)
-            σ = sum(δ)
-            l = length(δ) - 1
-            if σ > 0
-                closeness[u] = l / σ
-                if normalize
-                    n = l * 1.0 / (n_v - 1)
-                    closeness[u] *= n
-                end
-            else
-                closeness[u] = 0.0
-            end
-        end
-    end
-    return sdata(closeness)
+function distr_closeness_centrality(args...; kwargs...)
+    return error(
+        "`parallel = :distributed` requested, but SharedArrays or Distributed is not loaded"
+    )
 end
 
 function threaded_closeness_centrality(

src/Parallel/centrality/radiality.jl

@@ -1,27 +1,15 @@
-function radiality_centrality(g::AbstractGraph; parallel=:distributed)
+function radiality_centrality(g::AbstractGraph; parallel=:threads)
     return if parallel == :distributed
         distr_radiality_centrality(g)
     else
         threaded_radiality_centrality(g)
     end
 end
 
-function distr_radiality_centrality(g::AbstractGraph)::Vector{Float64}
-    n_v = nv(g)
-    vs = vertices(g)
-    n = ne(g)
-    meandists = SharedVector{Float64}(Int(n_v))
-    maxdists = SharedVector{Float64}(Int(n_v))
-
-    @sync @distributed for i in 1:n_v
-        d = Graphs.dijkstra_shortest_paths(g, vs[i])
-        maxdists[i] = maximum(d.dists)
-        meandists[i] = sum(d.dists) / (n_v - 1)
-        nothing
-    end
-    dmtr = maximum(maxdists)
-    radialities = collect(meandists)
-    return ((dmtr + 1) .- radialities) ./ dmtr
+function distr_radiality_centrality(args...; kwargs...)
+    return error(
+        "`parallel = :distributed` requested, but SharedArrays or Distributed is not loaded"
+    )
 end
 
 function threaded_radiality_centrality(g::AbstractGraph)::Vector{Float64}