🐛 fix disconnected communities (fixes #2)

Ensure that the aggregated graph is undirected. Since we are use low-memory scans for aggregate,
the aggregated graph is not guaranteed to be undirected by default, even in the input graph is undirected.

Author: wolfram77

inc/csr.hxx

@@ -25,6 +25,47 @@ inline K csrDegree(const vector<O>& offsets, K u) {
 
 
 
+#pragma region HAS EDGE
+/**
+ * Check if an edge exists in the graph.
+ * @param offsets offsets of the outgoing edges of vertices
+ * @param edgeKeys vertex ids of the outgoing edges of each vertex
+ * @param u source vertex id
+ * @param v target vertex id
+ * @returns true if edge exists, false otherwise
+ */
+template <class O, class K>
+inline bool csrHasEdge(const vector<O>& offsets, const vector<K>& edgeKeys, K u, K v) {
+  O i = offsets[u];
+  O I = offsets[u+1];
+  for (; i<I; ++i)
+    if (edgeKeys[i]==v) return true;
+  return false;
+}
+
+
+/**
+ * Check if an edge exists in the graph.
+ * @param offsets offsets of the outgoing edges of vertices
+ * @param degrees degree of each vertex
+ * @param edgeKeys vertex ids of the outgoing edges of each vertex
+ * @param u source vertex id
+ * @param v target vertex id
+ * @returns true if edge exists, false otherwise
+ */
+template <class O, class K>
+inline bool csrHasEdge(const vector<O>& offsets, const vector<K>& degrees, const vector<K>& edgeKeys, K u, K v) {
+  O i = offsets[u];
+  O I = offsets[u] + degrees[u];
+  for (; i<I; ++i)
+    if (edgeKeys[i]==v) return true;
+  return false;
+}
+#pragma endregion
+
+
+
+
 #pragma region FOREACH
 /**
  * Iterate over the target vertex ids of a source vertex in the graph.
@@ -340,15 +381,17 @@ inline void csrClearOmpW(vector<O>& offsets, vector<K>& degrees) {
 
 /**
  * Add an edge to the graph.
+ * @tparam CHECK check if edge already exists?
  * @param degrees degree of each vertex
  * @param edgeKeys vertex ids of the outgoing edges of each vertex
  * @param offsets offsets of the outgoing edges of vertices
  * @param u source vertex id
  * @param v target vertex id
  * @note Does not check if the edge already exists, or is there is available space.
  */
-template <class O, class K>
+template <bool CHECK=false, class O, class K>
 inline void csrAddEdgeU(vector<K>& degrees, vector<K>& edgeKeys, const vector<O>& offsets, K u, K v) {
+  if (CHECK && csrHasEdge(offsets, degrees, edgeKeys, u, v)) return;
   O n = degrees[u]++;
   O i = offsets[u] + n;
   edgeKeys[i] = v;
@@ -357,26 +400,30 @@ inline void csrAddEdgeU(vector<K>& degrees, vector<K>& edgeKeys, const vector<O>
 #ifdef OPENMP
 /**
  * Add an edge to the graph.
+ * @tparam CHECK check if edge already exists?
  * @param degrees degree of each vertex
  * @param edgeKeys vertex ids of the outgoing edges of each vertex
  * @param offsets offsets of the outgoing edges of vertices
  * @param u source vertex id
  * @param v target vertex id
  * @note Does not check if the edge already exists, or is there is available space.
  */
-template <class O, class K>
+template <bool CHECK=false, class O, class K>
 inline void csrAddEdgeOmpU(vector<K>& degrees, vector<K>& edgeKeys, const vector<O>& offsets, K u, K v) {
+  if (CHECK && csrHasEdge(offsets, degrees, edgeKeys, u, v)) return;
   O n = 0;
   #pragma omp atomic capture
   { n = degrees[u]; ++degrees[u]; }
   O i = offsets[u] + n;
+  if (CHECK && offsets[u+1]<=i) return;  // Check if there is available space.
   edgeKeys[i] = v;
 }
 #endif
 
 
 /**
  * Add a weighted edge to the graph.
+ * @tparam CHECK check if edge already exists?
  * @param degrees degree of each vertex
  * @param edgeKeys vertex ids of the outgoing edges of each vertex
  * @param edgeValues edge values of the outgoing edges of each vertex
@@ -386,8 +433,9 @@ inline void csrAddEdgeOmpU(vector<K>& degrees, vector<K>& edgeKeys, const vector
  * @param w associated weight of the edge
  * @note Does not check if the edge already exists, or is there is available space.
  */
-template <class O, class K, class E>
+template <bool CHECK=false, class O, class K, class E>
 inline void csrAddEdgeU(vector<K>& degrees, vector<K>& edgeKeys, vector<E>& edgeValues, const vector<O>& offsets, K u, K v, E w) {
+  if (CHECK && csrHasEdge(offsets, degrees, edgeKeys, u, v)) return;
   O n = degrees[u]++;
   O i = offsets[u] + n;
   edgeKeys[i]   = v;
@@ -397,6 +445,7 @@ inline void csrAddEdgeU(vector<K>& degrees, vector<K>& edgeKeys, vector<E>& edge
 #ifdef OPENMP
 /**
  * Add a weighted edge to the graph.
+ * @tparam CHECK check if edge already exists?
  * @param degrees degree of each vertex
  * @param edgeKeys vertex ids of the outgoing edges of each vertex
  * @param edgeValues edge values of the outgoing edges of each vertex
@@ -406,12 +455,14 @@ inline void csrAddEdgeU(vector<K>& degrees, vector<K>& edgeKeys, vector<E>& edge
  * @param w associated weight of the edge
  * @note Does not check if the edge already exists, or is there is available space.
  */
-template <class O, class K, class E>
+template <bool CHECK=false, class O, class K, class E>
 inline void csrAddEdgeOmpU(vector<K>& degrees, vector<K>& edgeKeys, vector<E>& edgeValues, const vector<O>& offsets, K u, K v, E w) {
+  if (CHECK && csrHasEdge(offsets, degrees, edgeKeys, u, v)) return;
   O n = 0;
   #pragma omp atomic capture
   { n = degrees[u]; ++degrees[u]; }
   O i = offsets[u] + n;
+  if (CHECK && offsets[u+1]<=i) return;  // Check if there is available space.
   edgeKeys[i]   = v;
   edgeValues[i] = w;
 }

inc/leidenLowmem.hxx

@@ -383,9 +383,14 @@ inline void leidenLowmemAggregateEdgesOmpW(vector<K>& ydeg, vector<K>& yedg, vec
       K d = (*mcs[t])[i];
       V w = (*mws[t])[i];
       if (!w) continue;
-      csrAddEdgeU(ydeg, yedg, ywei, yoff, c, d, W(w));
+      csrAddEdgeOmpU<true>(ydeg, yedg, ywei, yoff, c, d, W(w));
+      csrAddEdgeOmpU<true>(ydeg, yedg, ywei, yoff, d, c, W(w));
     }
   }
+  // Ensure `ydeg` does not exceed bounds.
+  #pragma omp parallel for schedule(auto)
+  for (K c=0; c<C; ++c)
+    ydeg[c] = min(ydeg[c], K(yoff[c+1] - yoff[c]));
 }
 
 

main.cxx

@@ -52,9 +52,11 @@ void runExperiment(const G& x) {
   // Follow a specific result logging format, which can be easily parsed later.
   auto flog = [&](const auto& ans, const char *technique, size_t numSlots=0) {
     printf(
-      "{%09.1fms, %09.1fms mark, %09.1fms init, %09.1fms firstpass, %09.1fms locmove, %09.1fms refine, %09.1fms aggr, %.3e slots, %04d iters, %03d passes, %01.9f modularity} %s\n",
+      "{%09.1fms, %09.1fms mark, %09.1fms init, %09.1fms firstpass, %09.1fms locmove, %09.1fms refine, %09.1fms aggr, %.3e slots, %04d iters, %03d passes, %01.9f modularity, %zu/%zu disconnected} %s\n",
       ans.time, ans.markingTime, ans.initializationTime, ans.firstPassTime, ans.localMoveTime, ans.refinementTime, ans.aggregationTime,
-      double(numSlots), ans.iterations, ans.passes, getModularity(x, ans, M), technique
+      double(numSlots), ans.iterations, ans.passes, getModularity(x, ans, M),
+      countValue(communitiesDisconnectedOmp(x, ans.membership), char(1)),
+      communities(x, ans.membership).size(), technique
     );
   };
   // Get community memberships on original graph (static).