Added traces to track where the time is spent during edge generation

Author: ArnauPrat

src/main/java/ldbc/snb/datagen/generator/ClusteringKnowsGenerator.java

@@ -508,34 +508,48 @@ void fillGraphWithRemainingEdges(ClusteringInfo cInfo, ArrayList<Community> comm
 
     public void generateKnows( ArrayList<Person> persons, int seed, ArrayList<Float> percentages, int step_index )  {
 
+        long start, end;
         rand.setSeed(seed);
         this.percentages = percentages;
         this.stepIndex = step_index;
 
+        start = System.currentTimeMillis();
         ArrayList<Community>  communities = generateCommunities(persons);
+        end = System.currentTimeMillis();
+        System.out.println("Time to configure communities: "+(end-start));
+
         ClusteringInfo cInfo = new ClusteringInfo( persons.size(), communities );
 
         System.out.println("Number of generated communities: "+communities.size());
 
 
+        start = System.currentTimeMillis();
         for( Community c : communities ) {
             c.p_ = 1.0f;
             computeCommunityInfo(cInfo, c, 1.0f);
         }
+        end = System.currentTimeMillis();
+        System.out.println("Time to compute initial community information: "+(end-start));
 
+        start = System.currentTimeMillis();
         for( Community c : communities ) {
             c.p_ = 1.0f;
             estimateCCCommunity(cInfo, c, c.p_);
         }
 
         float maxCC = clusteringCoefficient(communities, cInfo);
+        end = System.currentTimeMillis();
         System.out.println("maxCC: "+maxCC);
+        System.out.println("Time to compute maximum CC: "+(end-start));
 
+        start = System.currentTimeMillis();
         for( Community c : communities ) {
             c.p_ = 0.5f;//rand.nextFloat();
             //c.p_ = rand.nextFloat();
             estimateCCCommunity(cInfo, c, c.p_ );
         }
+        end = System.currentTimeMillis();
+        System.out.println("Time to compute the initial solution: "+(end-start));
 
         PersonGraph graph;
         boolean iterate;
@@ -544,15 +558,21 @@ public void generateKnows( ArrayList<Person> persons, int seed, ArrayList<Float>
         do {
             System.out.println("Starting refinement iteration");
             iterate = false;
+            start = System.currentTimeMillis();
             refineCommunities(cInfo, communities, fakeTargetCC);
+            end = System.currentTimeMillis();
+            System.out.println("Time to refine communities: "+(end-start));
+
             System.out.println("Creating graph");
 
-            long start, end;
+            start = System.currentTimeMillis();
             for(Community c : communities ) {
                 createEdgesCommunityCore(persons, c);
                 createEdgesCommunityPeriphery(cInfo, persons, c);
             }
             fillGraphWithRemainingEdges(cInfo, communities, persons);
+            end = System.currentTimeMillis();
+            System.out.println("Time to generate graph: "+(end-start));
 
             graph = new PersonGraph(persons);
             System.out.println("Computing clustering coefficient");

src/main/java/ldbc/snb/datagen/generator/LDBCDatagen.java

@@ -268,8 +268,8 @@ public int runGenerateJob(Configuration conf) throws Exception {
                 + " total seconds");
         System.out.println("Person generation time: "+((endPerson - startPerson) / 1000));
         System.out.println("University correlated edge generation time: "+((endUniversity - startUniversity) / 1000));
-        //System.out.println("Interest correlated edge generation time: "+((endInterest - startInterest) / 1000));
-        //System.out.println("Random correlated edge generation time: "+((endRandom - startRandom) / 1000));
+        System.out.println("Interest correlated edge generation time: "+((endInterest - startInterest) / 1000));
+        System.out.println("Random correlated edge generation time: "+((endRandom - startRandom) / 1000));
         System.out.println("Edges merge time: "+((endMerge - startMerge) / 1000));
         System.out.println("Person serialization time: "+((endPersonSerializing - startPersonSerializing) / 1000));
         System.out.println("Person activity generation and serialization time: "+((endPersonActivity - startPersonActivity) / 1000));

src/main/java/ldbc/snb/datagen/generator/distribution/MOEZipfDistribution.java

@@ -21,7 +21,7 @@ public class MOEZipfDistribution implements DegreeDistribution  {
     public void initialize(Configuration conf) {
         ALPHA_ = conf.getDouble("ldbc.snb.datagen.generator.distribution.MOEZipfDistribution.alpha",ALPHA_);
         DELTA_ = conf.getDouble("ldbc.snb.datagen.generator.distribution.MOEZipfDistribution.delta",DELTA_);
-        zipf_ = new org.apache.commons.math3.distribution.ZipfDistribution(10000, ALPHA_);
+        zipf_ = new org.apache.commons.math3.distribution.ZipfDistribution(5000, ALPHA_);
         random_ = new Random();
     }