Dune cdr the Single-Phase Protodune
Yüklə 4,82 Kb. Pdf görüntüsü
|
Chapter 5: Software and Computing 5–136 a figure of merit, for charged pion events with two values of test beam momentum, 1 GeV/c and 4 GeV/c. Figure 5.2: CNN-based selection of the electromagnetic component of events: deconvoluted ADC waveforms used as the input to the algorithm (left); selected electromagnetic (blue) and track-like (red) activity (right). Positive pion in the ProtoDUNE test beam at 2.5 GeV/c is shown (the particle track incoming from the left). All parts are correctly recognized in this event. Figure 5.3: Summed hit ADC areas for hits selected by the CNN as originating from electromagnetic activity versus the summed hit ADC areas for hits selected with MC truth information as originating from electron tracks. This comparison is shown for simulated ProtoDUNE-SP events with a π + with a momentum of 1 GeV/c (left) and 4 GeV/c (right). EMShower The EMShower package [43] takes the output of the Blurred Clustering Algorithm and produces energies, angles, and start positions for 3D showers, as well as the dE/dx in the initial part of the shower. Identifying events with two showers consistent with π 0 → γγ decays allows for an in situ calibration of the electromagnetic energy scale as well as the performance of shower identification and reconstruction for photons that are produced inside the detector. A distribution of reconstructed π 0 masses in Monte Carlo is shown in Figure 5.4. ProtoDUNE Single-Phase Technical Design Report Chapter 5: Software and Computing 5–137 ) 2 (GeV/c 0 π Reconstructed m 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Events 0 100 200 300 400 500 600 700 Figure 5.4: The reconstructed invariant masses of π 0 candidates in Monte Carlo using the BlurredCluster and EMShower algorithms. PANDORA The reconstruction framework PANDORA [44] also works by building up a 3D picture from 2D re- constructed objects. PANDORA is a flexible framework developed for International Linear Collider (ILC) detector simulation, and provides a convenient way to develop algorithms for reconstructing particles. In all, more than 80 algorithms, each targeting a specific topology, have been incor- porated into PANDORA to date. PANDORA allows multiple reconstruction passes through the data. Different criteria for clustering hits into tracks and showers may be applied when seeking to remove cosmic rays than when identifying signal events. PANDORA follows a process that clusters hits in 2D, reconstructs vertices in 3D, reconstructs tracks in 3D, reconstructs showers in 3D, performs a mop-up step in 2D and 3D, and finally performs full event-building in 3D. Plots of the efficiency, the completeness, and the difference between the true and reconstructed track lengths for single muons with momentum between 300 MeV and 5 GeV in the ProtoDUNE-SP geometry are shown in Figure 5.5. The PANDORA algorithm performs very well, although a small inefficiency occurs in the current algorithm’s ability to match track segments from one APA’s drift volume to another. PMA Another approach to 3D reconstruction in LArTPC detectors is referred to as the Projection Match- ing Algorithm (PMA) [45]. PMA was primarily developed as a technique for 3D reconstruction of individual particle trajectories (trajectory fits). Instead of building up a 3D hypothesis from ProtoDUNE Single-Phase Technical Design Report Chapter 5: Software and Computing 5–138 Muon Momentum (GeV) 0 1 2 3 4 5 Tracking Efficiency 0 0.2 0.4 0.6 0.8 1 Track Completeness 0 0.2 0.4 0.6 0.8 1 1.2 0 500 1000 1500 2000 2500 3000 3500 4000 h_Ecomplet_lepton Entries 9314 Mean 0.8614 RMS 0.217 Ecomplet Lepton True length - Reco length (cm) 100 − 80 − 60 − 40 − 20 − 0 20 40 60 80 100 0 200 400 600 800 1000 1200 1400 1600 h_trackRes_lepton Entries 9314 Mean 19.92 RMS 32.68 Muon Residual Figure 5.5: Performance of the PANDORA reconstruction algorithm for single muons with momentum between 300 MeV/c and 5 GeV/c. The top left-hand figure shows the tracking efficiency as a function of muon momentum, the top right-hand figure shows the distribution of tracking completeness, and the bottom figure shows the distribution of the difference between the reconstructed muon track length and the true track length. ProtoDUNE Single-Phase Technical Design Report Chapter 5: Software and Computing 5–139 2D clusters, it starts with the 3D hypothesis and compares the 2D projection of the predicted trajectory of a particle with the observed data. Association of hits between the 2D planes is not needed in this approach, improving its performance in problematic cases, such as isochronous and short tracks. PMA can take as input the output from different pattern recognition algorithms, from LineClus- ter [41] to WireCell (described below). Because these 2D algorithms are run on each 2D projection independently, and because of detector defects, clusters from particles may be broken into several smaller pieces, fractions of 2D clusters may be missing, and clusters obtained from complementary projections are not guaranteed to cover corresponding sections of trajectories. Such behavior is expected since ambiguous trajectories can be resolved only if the information from multiple 2D projections is used. PMA performs higher-level pattern recognition using as input clustering infor- mation from all projections in order to search for the best matching combinations of clusters. The algorithm also attempts to correct hit-to-cluster assignments using properties of 3D reconstructed objects. Plots of the efficiency, the completeness, and the difference between the true and reconstructed track lengths for single muons with momentum between 300 MeV and 5 GeV in the ProtoDUNE-SP geometry are shown in Figure 5.6. Muon Momentum (GeV) 0 1 2 3 4 5 Tracking Efficiency 0 0.2 0.4 0.6 0.8 1 Track Completeness 0 0.2 0.4 0.6 0.8 1 1.2 0 1000 2000 3000 4000 5000 6000 h_Ecomplet_lepton Entries 9600 Mean 0.9566 RMS 0.06367 Ecomplet Lepton True length - Reco length (cm) 100 − 80 − 60 − 40 − 20 − 0 20 40 60 80 100 0 1000 2000 3000 4000 5000 h_trackRes_lepton Entries 9600 Mean 2.189 RMS 10.87 Muon Residual Figure 5.6: Performance of the PMA reconstruction algorithm for single muons with momentum between 300 MeV/c and 5 GeV/c. The left-hand figure shows the tracking efficiency as a function of muon momentum, the middle figure shows the distribution of tracking completeness, and the right-hand figure shows the distribution of the difference between the reconstructed muon track length and the true track length. PMA has been used successfully in ProtoDUNE-SP to reconstruct simulated beam particles and cosmic muons [46]. In order to illustrate the performance of the entire reconstruction chain, ProtoDUNE Single-Phase Technical Design Report Yüklə 4,82 Kb. Dostları ilə paylaş:
|