setIntersectionBenchmark.java

  static int setIntersectionBenchmark() throws java.io.IOException {
    final int nBags        = 1000;
    final int maxBagLength = 10;
    
    Scanner                     scanner    = new Scanner(new File("10000ran.dat"));
    ArrayList<HashSet<Integer>> bags       = new ArrayList<HashSet<Integer>>(nBags); 
    int                         similarity = 0;
    
    scanner.useDelimiter(",\\s*");
    
    for (int iBag = 0; iBag < nBags; ++iBag) {
      HashSet<Integer> bag = new HashSet<Integer>(maxBagLength);
    
      for (int j = 0; j < maxBagLength; ++j)
        bag.add(round(scanner.nextFloat()*maxBagLength));
      
      bags.add(bag);
    }

    scanner.close();

    for (int i = 0; i < nBags; ++i)
      for (int j = i+1; j < nBags; ++j) {
        HashSet<Integer> s = new HashSet<Integer>(bags.get(i));
        
        s.retainAll(bags.get(j));
        similarity += s.size();
      }
    
    return similarity;    
  }