setIntersectionBenchmark.cpp

int setIntersectionBenchmark() 
{
  const int nBags        = 1000;
  const int maxBagLength = 10;
    
  vector<set<int>> bags;
  ifstream         input("10000ran.dat");  // http://www.agner.org/random/10000ran.zip
  int              similarity = 0;
    
  bags.reserve(nBags);

  for (int iBag = 0; iBag < nBags; ++iBag) {
    bags.push_back(set<int>());
    
    for (int j = 0; j < maxBagLength; ++j) {
      string number;

      getline(input, number, ',');
      bags.back().insert(round(atof(number.c_str())*maxBagLength));
    }
  }
    
  for (int i = 0; i < bags.size(); ++i)
    for (int j = i+1; j < bags.size(); ++j) {
      set<int> intersec;

      set_intersection(bags[i].begin(), bags[i].end(), bags[j].begin(), bags[j].end(), inserter(intersec, intersec.begin()));
      similarity += intersec.size();
    }

  return similarity;    
}