BinningTools/src/ClusterizingHistogram.cc

0001 #include "Utilities/BinningTools/interface/ClusterizingHistogram.h"
0002 #include <iostream>
0003
0004 using namespace std;
0005
0006 ClusterizingHistogram::ClusterizingHistogram(int nb, float xmi, float xma)
0007     : my_nbins(nb), xmin(xmi), xmax(xma), my_entries(0), my_underflows(0), my_overflows(0) {
0008   bin_entries = new int[my_nbins];
0009   bin_means = new float[my_nbins];
0010   binsiz = (xmax - xmin) / my_nbins;
0011   for (int i = 0; i < my_nbins; i++) {
0012     bin_entries[i] = 0;
0013     bin_means[i] = 0.;
0014   }
0015 }
0016 ClusterizingHistogram::~ClusterizingHistogram() {
0017   delete[] bin_entries;
0018   delete[] bin_means;
0019 }
0020
0021 int ClusterizingHistogram::bin(float x) const {
0022   if (x < xmin)
0023     return -1;
0024   else if (x > xmax)
0025     return my_nbins;
0026   else
0027     return int((x - xmin) / binsiz);
0028 }
0029 int ClusterizingHistogram::bin(double x) const {
0030   if (x < xmin)
0031     return -1;
0032   else if (x > xmax)
0033     return my_nbins;
0034   else
0035     return int((x - xmin) / binsiz);
0036 }
0037
0038 void ClusterizingHistogram::fill(float x) {
0039   if (x < xmin)
0040     my_underflows++;
0041   else if (x > xmax)
0042     my_overflows++;
0043   else {
0044     int bin = int((x - xmin) / binsiz);
0045     if (bin > my_nbins - 1)
0046       bin = my_nbins - 1;
0047     ++bin_entries[bin];
0048     bin_means[bin] += x;
0049     my_entries++;
0050     // may be problematic for negative x; check!
0051   }
0052 }
0053
0054 vector<float> ClusterizingHistogram::clusterize(float resol) {
0055   vector<float> clust;
0056   int nclust = 0;
0057   bool inclust = false;
0058   float last_pos = xmin - 1000. * resol;
0059   int sum = 0;
0060   float sumx = 0;
0061   for (int i = 0; i < my_nbins; i++) {
0062     if (bin_entries[i] != 0) {
0063       if (fabs(bin_pos(i) - last_pos) > resol) {
0064         inclust = false;
0065         if (nclust != 0)
0066           clust.push_back(sumx / sum);  // create cluster
0067       }
0068       if (!inclust) {
0069         nclust++;
0070         sumx = 0.;
0071         sum = 0;
0072       }
0073       sum += bin_entries[i];
0074       sumx += bin_means[i];
0075       last_pos = bin_pos(i);
0076       inclust = true;
0077     }
0078   }
0079   if (nclust != 0)
0080     clust.push_back(sumx / sum);  // create last cluster
0081   return clust;
0082 }
0083
0084 void ClusterizingHistogram::dump() const { dump(0, my_nbins); }
0085
0086 void ClusterizingHistogram::dump(int i1, int i2) const {
0087   cout << "Dumping ClusterizingHistogram contents:" << endl;
0088   for (int i = max(i1, 0); i < min(i2, my_nbins); i++) {
0089     cout << i << "  " << bin_entries[i] << "   " << bin_pos(i) << endl;
0090   }
0091   cout << "Underflows: " << my_underflows << endl;
0092   cout << "Overflows:  " << my_overflows << endl;
0093   cout << "Total number of entries: " << my_entries << endl;
0094 }
0095
0096 void ClusterizingHistogram::dump(float x1, float x2) const { dump(bin(x1), bin(x2)); }
0097 void ClusterizingHistogram::dump(double x1, double x2) const { dump(bin(x1), bin(x2)); }
0098 void ClusterizingHistogram::dump(float x1, double x2) const { dump(bin(x1), bin(x2)); }
0099 void ClusterizingHistogram::dump(double x1, float x2) const { dump(bin(x1), bin(x2)); }
0100
0101 void ClusterizingHistogram::reset() {
0102   my_entries = 0;
0103   my_underflows = 0;
0104   my_overflows = 0;
0105   for (int i = 0; i < my_nbins; i++) {
0106     bin_entries[i] = 0;
0107     bin_means[i] = 0.;
0108   }
0109 }