Changed DegreeDistribution class hierarchy to support non bucketized distributions

Author: ArnauPrat

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

@@ -24,8 +24,8 @@ public class PersonGenerator {
 
     public PersonGenerator( Configuration conf, String degreeDistribution ) {
 	    try{
-		    degreeDistribution_ = new DegreeDistribution();
-            degreeDistribution_.initialize(conf, (BucketedDistribution) Class.forName(degreeDistribution).newInstance());
+		    degreeDistribution_ = (DegreeDistribution)Class.forName(degreeDistribution).newInstance();
+            degreeDistribution_.initialize(conf);
 	    } catch(ClassNotFoundException e) {
 		    System.out.print(e.getMessage());
             } catch(IllegalAccessException e) {

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

@@ -10,7 +10,7 @@
 /**
  * Created by aprat on 26/02/15.
  */
-public class AltmannDistribution implements BucketedDistribution {
+public class AltmannDistribution extends BucketedDistribution {
 
     private int NUM_BUCKETS_ = 100;
     private int POPULATION_ = 10000;

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

@@ -16,36 +16,11 @@
  *
  * @author aprat
  */
-public class DegreeDistribution {
+public interface DegreeDistribution {
 
-    private ArrayList<Bucket> buckets_;
-    private ArrayList<Random> randomDegree_;
-    private Random randomPercentile_;
+	public void initialize( Configuration conf );
 
+    public void reset (long seed);
 
-	public void initialize( Configuration conf, BucketedDistribution dist ) {
-        dist.initialize(conf);
-        buckets_ = dist.getBuckets();
-        randomPercentile_ = new Random(0);
-        randomDegree_ = new ArrayList<Random>();
-        for (int i = 0; i < buckets_.size(); i++) {
-            randomDegree_.add(new Random(0));
-        }
-    }
-
-	public void reset(long seed) {
-        Random seedRandom = new Random(53223436L + 1234567 * seed);
-        for (int i = 0; i < buckets_.size(); i++) {
-            randomDegree_.get(i).setSeed(seedRandom.nextLong());
-        }
-        randomPercentile_.setSeed(seedRandom.nextLong());
-    }
-
-	public long nextDegree() {
-        int idx = randomPercentile_.nextInt(buckets_.size());
-        int minRange = (int)(buckets_.get(idx).min());
-        int maxRange = (int)(Math.floor(buckets_.get(idx).max()));
-        if( maxRange < minRange ) maxRange = minRange;
-        return randomDegree_.get(idx).nextInt( maxRange - minRange  + 1) + minRange;
-    }
+	public long nextDegree();
 }

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

@@ -60,7 +60,7 @@
  * 2) Randomly select a social degree in the range of that bucket
  */
 
-public class FacebookDegreeDistribution implements BucketedDistribution {
+public class FacebookDegreeDistribution extends BucketedDistribution {
     private int mean_ = 0;
     private static final int FB_MEAN_ = 190;
     private static final int BUCKET_NUM_ = 100;
@@ -71,15 +71,12 @@ public FacebookDegreeDistribution() {
 
     @Override
     public ArrayList<Bucket> getBuckets() {
-        return buckets_;
-    }
-
-    public void initialize( Configuration conf) {
         mean_ = (int) Math.round(Math.pow(DatagenParams.numPersons, (0.512 - 0.028 * Math.log10(DatagenParams.numPersons))));
         System.out.println("Mean = " + mean_);
         buckets_ = new ArrayList<Bucket>();
         loadFBBuckets();
         rebuildBucketRange();
+        return buckets_;
     }
 
     public void loadFBBuckets() {

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

@@ -5,32 +5,27 @@
         import ldbc.snb.datagen.generator.distribution.utils.BucketedDistribution;
         import org.apache.commons.math3.distribution.GeometricDistribution;
         import org.apache.hadoop.conf.Configuration;
+        import org.apache.hadoop.hdfs.server.namenode.FSImageFormatProtobuf;
 
         import java.util.ArrayList;
 
 /**
  * Created by aprat on 5/03/15.
  */
-public class GeoDistribution implements BucketedDistribution {
+public class GeoDistribution implements DegreeDistribution {
 
-    private ArrayList<Bucket> buckets_;
     private GeometricDistribution geo_;
     private double ALPHA_ = 0.12;
 
-    @Override
-    public ArrayList<Bucket> getBuckets() {
-        return buckets_;
-    }
-
-    @Override
     public void initialize(Configuration conf) {
         geo_ = new GeometricDistribution(ALPHA_);
+    }
 
-        ArrayList<Double> histogram = new ArrayList<Double>();
-        for( int i = 1; i <= DatagenParams.numPersons; ++i ) {
-            histogram.add(DatagenParams.numPersons * geo_.probability(i));
-        }
+    public void reset (long seed) {
+        geo_.reseedRandomGenerator(seed);
+    }
 
-        buckets_ = Bucket.bucketizeHistogram(histogram,100);
+    public long nextDegree(){
+        return geo_.sample();
     }
 }

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

@@ -10,7 +10,7 @@
 /**
  * Created by aprat on 4/03/15.
  */
-public class MOEZipfDistribution implements BucketedDistribution {
+public class MOEZipfDistribution extends BucketedDistribution {
 
     private int NUM_BUCKETS_ = 100;
     private int POPULATION_ = 10000;

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

@@ -11,7 +11,7 @@
 /**
  * Created by aprat on 5/03/15.
  */
-public class WeibullDistribution implements BucketedDistribution {
+public class WeibullDistribution extends BucketedDistribution {
 
     private ArrayList<Bucket> buckets_;
     private org.apache.commons.math3.distribution.WeibullDistribution weibull_;
@@ -29,6 +29,7 @@ public void initialize(Configuration conf) {
 
         ArrayList<Double> histogram = new ArrayList<Double>();
         for( int i = 1; i <= DatagenParams.numPersons; ++i ) {
+            System.out.println(weibull_.probability(i));
             histogram.add(DatagenParams.numPersons * weibull_.probability(i));
         }
 

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

@@ -11,7 +11,7 @@
 /**
  * Created by aprat on 5/03/15.
  */
-public class ZetaDistribution implements BucketedDistribution {
+public class ZetaDistribution extends BucketedDistribution {
 
     private ArrayList<Bucket> buckets_;
     private ZipfDistribution zipf_;

src/main/java/ldbc/snb/datagen/generator/distribution/utils/BucketedDistribution.java

@@ -1,15 +1,44 @@
 package ldbc.snb.datagen.generator.distribution.utils;
 
+import ldbc.snb.datagen.generator.distribution.DegreeDistribution;
 import org.apache.hadoop.conf.Configuration;
 
 import java.util.ArrayList;
+import java.util.Random;
 
 /**
  * Created by aprat on 3/03/15.
  */
-public interface BucketedDistribution {
+public abstract class BucketedDistribution implements DegreeDistribution {
 
-    public ArrayList<Bucket> getBuckets();
+    private ArrayList<Bucket> buckets_;
+    private ArrayList<Random> randomDegree_;
+    private Random randomPercentile_;
 
-    public void initialize( Configuration conf );
+    public abstract ArrayList<Bucket> getBuckets();
+
+    public void initialize( Configuration conf ) {
+        buckets_ = this.getBuckets();
+        randomPercentile_ = new Random(0);
+        randomDegree_ = new ArrayList<Random>();
+        for (int i = 0; i < buckets_.size(); i++) {
+            randomDegree_.add(new Random(0));
+        }
+    };
+
+    public void reset(long seed) {
+        Random seedRandom = new Random(53223436L + 1234567 * seed);
+        for (int i = 0; i < buckets_.size(); i++) {
+            randomDegree_.get(i).setSeed(seedRandom.nextLong());
+        }
+        randomPercentile_.setSeed(seedRandom.nextLong());
+    }
+
+    public long nextDegree() {
+        int idx = randomPercentile_.nextInt(buckets_.size());
+        int minRange = (int)(buckets_.get(idx).min());
+        int maxRange = (int)(Math.floor(buckets_.get(idx).max()));
+        if( maxRange < minRange ) maxRange = minRange;
+        return randomDegree_.get(idx).nextInt( maxRange - minRange  + 1) + minRange;
+    }
 }