passing edges through an unordered map first to remove dupes

GregorySchwing · Aug 12, 2021 · acd4bdc · acd4bdc
1 parent f24cf9c
commit acd4bdc
Show file tree

Hide file tree

Showing 2 changed files with 41 additions and 15 deletions.
diff --git a/simpleParallel/COO.cpp b/simpleParallel/COO.cpp
@@ -48,32 +48,35 @@ void COO::BuildCOOFromFile(std::string filename){
     for(auto& row: CSVRange(file, sep))
     {
         //std::cout << "adding (" << std::stoi(row[0],&sz) 
-        //<< ", " << std::stoi(row[1],&sz) << ")" << std::endl; 
-        addEdgeSymmetric(std::stoi(row[0],&sz), 
+        //<< ", " << std::stoi(row[1],&sz) << ")" << std::endl;
+        addEdgeToMap(std::stoi(row[0],&sz), 
                         std::stoi(row[1],&sz), 1);
         //coordinateFormat->addEdgeSimple(std::stoi(row[0],&sz), 
         //                                    std::stoi(row[1],&sz), 1);
     }
 
+    for (auto & edge : um)
+        addEdgeSymmetric(edge.first.first, edge.first.second, 1);
+
     size = new_values.size();
     // vlog(e)
     sortMyself();
 }
 
 void COO::BuildTheExampleCOO(){
-    addEdgeSymmetric(0,1,1);
-    addEdgeSymmetric(0,4,1);
-    addEdgeSymmetric(1,4,1);
-    addEdgeSymmetric(1,5,1);
-    addEdgeSymmetric(1,6,1);
-    addEdgeSymmetric(2,4,1);
-    addEdgeSymmetric(2,6,1);
-    addEdgeSymmetric(3,5,1);
-    addEdgeSymmetric(3,6,1);
-    addEdgeSymmetric(4,7,1);
-    addEdgeSymmetric(4,8,1);
-    addEdgeSymmetric(5,8,1);
-    addEdgeSymmetric(6,9,1);
+    addEdgeToMap(0,1,1);
+    addEdgeToMap(0,4,1);
+    addEdgeToMap(1,4,1);
+    addEdgeToMap(1,5,1);
+    addEdgeToMap(1,6,1);
+    addEdgeToMap(2,4,1);
+    addEdgeToMap(2,6,1);
+    addEdgeToMap(3,5,1);
+    addEdgeToMap(3,6,1);
+    addEdgeToMap(4,7,1);
+    addEdgeToMap(4,8,1);
+    addEdgeToMap(5,8,1);
+    addEdgeToMap(6,9,1);
 
     size = new_values.size();
     // vlog(e)
@@ -151,6 +154,15 @@ void COO::addEdgeASymmetric(int u, int v, int weight){
 }
 
 
+/* Works but the amount of data is 2x */
+void COO::addEdgeToMap(int u, int v, int weight){
+    std::pair<int, int> in(u, v);
+    std::pair<int, int> out(v, u);
+
+    um[in] = true;
+    um[out] = true;
+}
+
 /* Works but the amount of data is 2x */
 void COO::addEdgeSymmetric(int u, int v, int weight){
     new_row_indices.push_back(u);

diff --git a/simpleParallel/COO.h b/simpleParallel/COO.h
@@ -8,6 +8,18 @@
 #include "../common/CSVRange.h"
 #include  <iterator>
 #include <algorithm> /* rand */
+#include <unordered_map>
+
+// A hash function used to hash a pair of any kind
+struct hash_pair {
+    template <class T1, class T2>
+    size_t operator()(const std::pair<T1, T2>& p) const
+    {
+        auto hash1 = std::hash<T1>{}(p.first);
+        auto hash2 = std::hash<T2>{}(p.second);
+        return hash1 ^ hash2;
+    }
+};
 
 class COO final : public SparseMatrix
 {
@@ -29,6 +41,7 @@ class COO final : public SparseMatrix
     void addEdgeSimple(int u, int v, int weight);
     void addEdgeASymmetric(int u, int v, int weight);
     void addEdgeSymmetric(int u, int v, int weight);
+    void addEdgeToMap(int u, int v, int weight);
     int GetNumberOfVertices();
     int GetNumberOfEdges();
     void BuildTheExampleCOO();
@@ -39,6 +52,7 @@ class COO final : public SparseMatrix
     int vertexCount;
 
 private:
+std::unordered_map<std::pair<int, int>, bool, hash_pair> um;
 bool isSorted;
 };
 #endif