MTT0PatternRecognition.cxx

Go to the documentation of this file.

/*
  Copyright (C) 2002-2020 CERN for the benefit of the ATLAS collaboration
*/
 
#include "MdtCalibT0/MTT0PatternRecognition.h"
 
#include "AthenaKernel/getMessageSvc.h"
#include "GaudiKernel/MsgStream.h"
#include "TGraph.h"
#include "TH1.h"
#include "TLine.h"
 
namespace MuonCalib {

    // estimate_background(const TH1 &hist, double scale_min)   //
 
    bool MTT0PatternRecognition::estimate_background(TH1F* hist, double scale_min) {
        // Get first bin in input histogram which is not empty.     This avoids underestimation of the background rate if the range of
        // the input histogram exceeds the TDC-range
        int min(-1);
        for (int i = 1; i <= hist->GetNbinsX(); i++) {
            if (hist->GetBinContent(i) > 0) {
                min = i;
                break;
            }
        }
        if (min == -1) {
            MsgStream log(Athena::getMessageSvc(), "MTT0PattternRecognition");
            log << MSG::WARNING << "estimate_background() - No hits in input histogram!" << endmsg;
            m_error = true;
            return false;
        }
        double maxx = scale_min - 40;
        int max = hist->FindBin(maxx);
        // we want 10 bins minimum for the background estimate
        if (max - min < m_settings->MinBackgroundBins()) {
            min = max - 128;
            if (min < 0) min = 0;
        }
        // if there are still not enough bins for the background estimate we have a problem
        if (max - min < m_settings->MinBackgroundBins()) {
            MsgStream log(Athena::getMessageSvc(), "MTT0PattternRecognition");
            log << MSG::WARNING << "estimate_background() - Rising edge is to glose to lower histogram range!" << endmsg;
            m_error = true;
            return false;
        }
        // calculate average bin content
        m_background = 0.0;
        double back_squared = 0.0;
        double n_bins = 0.0;
        double referece_chi2 = 0.0;
        for (int i = min; i < max; i++) {
            n_bins++;
            m_background += hist->GetBinContent(i);
            back_squared += hist->GetBinContent(i) * hist->GetBinContent(i);
            if (n_bins == m_settings->MinBackgroundBins()) {
                double bac = m_background / n_bins;
                referece_chi2 = 2 * (back_squared / n_bins - bac * bac);
            }
            if (n_bins > m_settings->MinBackgroundBins()) {
                double bac = m_background / n_bins;
                double chi2 = 2 * (back_squared / n_bins - bac * bac);
                if (chi2 > 5 * referece_chi2) break;
            }
        }
        m_background /= n_bins;
        // store lower edge of fit range
        m_fit_min = hist->GetBinCenter(min);
        m_t0_est = 0.5 * (scale_min + hist->GetBinCenter(max));
        // mark selected range
        if (m_draw_debug_graph) {
            (new TLine(hist->GetBinCenter(min), 0, hist->GetBinCenter(min), hist->GetMaximum()))->Write("t0_back_left");
            (new TLine(hist->GetBinCenter(max), 0, hist->GetBinCenter(max), hist->GetMaximum()))->Write("t0_back_right");
        }
 
        return true;
    }

    // estimate_height(const TH1 &hist) //
 
    double MTT0PatternRecognition::estimate_height(TH1F* hist) {
        // smooth histogram
        if (!m_vbh.Smooth(hist->GetBinWidth(1))) {
            m_error = true;
            return -1;
        }
        if (!m_vbh.Smooth(hist->GetBinWidth(1) / 2)) {
            m_error = true;
            return -1;
        }
        // draw debug graphs
        if (m_draw_debug_graph) {
            m_vbh.DenistyGraph()->Write("t0_density");
            m_vbh.BinWidthGraph()->Write("t0_binwidth");
            m_vbh.BinContentGraph()->Write("t0_bin_content");
            m_vbh.DiffDensityGraph()->Write("t0_diff_density");
            m_vbh.DiffDensityGraph()->Write("t0_diffbinwidth");
        }
        // get the range in which the smallest, and second smallest bins are
        double lower = m_vbh.GetSortedBin(0).Width();
        double left(9e9), right(-9e9);
        //  bool smallest_found(false);
        for (unsigned int i = 0; i < m_vbh.GetNumberOfBins(); i++) {
            if (m_vbh.GetSortedBin(i).Left() < left) left = m_vbh.GetSortedBin(i).Left();
            if (m_vbh.GetSortedBin(i).Right() > right) right = m_vbh.GetSortedBin(i).Right();
            // NOTE: For greater numerical stability m_vbh.GetSortedBin(i).Width() is considered greater only if it is greater by
            // hist->GetBinWidth(1) / 100.0. I have seen differences in Binwidth in the order of 1e-13. On the other hand, due to the
            // binning of the input histogram, e real difference in the binwidth of the VBH n*hist->GetBinWidth(1).
            if ((m_vbh.GetSortedBin(i).Width() - lower) > hist->GetBinWidth(1) / 100.0) {
                // reqire minimum distance between bins, and minimum number if significant bins
                if (right - left > 50 && i > 20) break;
                lower = m_vbh.GetSortedBin(i).Width();
            }
        }
        // mark selected range
        if (m_draw_debug_graph) {
            (new TLine(left, 0, left, hist->GetMaximum()))->Write("t0_height_left");
            (new TLine(right, 0, right, hist->GetMaximum()))->Write("t0_height_right");
        }
        // calcularte mean bin content of input histogram in range
        int bleft(hist->FindBin(left)), bright(hist->FindBin(right));
        double nbins = 0.0;
        m_height = 0.0;
        for (int i = bleft; i <= bright; i++) {
            m_height += hist->GetBinContent(i);
            nbins++;
        }
        if (nbins < 1) {
            MsgStream log(Athena::getMessageSvc(), "MTT0PattternRecognition");
            log << MSG::WARNING << "estimate_height() - top region is too small! left=" << left << " right=" << right << endmsg;
            m_error = true;
            return -1;
        }
        m_height /= nbins;
        m_fit_max = right;
        return left;
    }
 
}  // namespace MuonCalib