Reconstruction Challenges in Large Volume Liquid Argon Detectors

Talk delivered at the Institute of Physics Nuclear & Particle Physics Divisional Conference 2011 (IoP NPPD11, Glasgow).

Transcript

1. Title Author Date Reconstruction Challenges In Large Volume Liquid Argon Detectors Andrew J. Bennieston IoP NPPD 2011, Glasgow
2. Liquid Argon TPCs Detectors on scale of 100 kton required for next generation ν experiments E Readout Charged Particle Ionisation Charge ! ! !"#$%&#'()*+,'$-.//&#$0"12/#) 3&/#)(*$4.5&,#&2,$670389:;<$=#.(),+*$>?@?A$ "((BACC,+D,'*E'#)+E'"C'*+F#)#+'#G,HB&1IEBIJ'*+F7DKLMNO@ A. Guglielmi (ICARUS), Neutrino 2010
3. The Challenges Automated reconstruction software not yet established (ICARUS, arXiv:0812.2373)
4. The Road To Analysis Experiment & Simulation Common Data Format Reconstruction Pipeline Clustering Performance Analysis USA (LArSoft): MicroBOONE ArgoNEUT LBNE Europe: Neutrino Factory IDS LAGUNA LBNO Japan: 250 litre chamber Calorimetry Tracking Particle ID
5. Reconstruction Algorithms Clustering DBSCAN OPTICS Cellular Automata conservative clustering Feature Extraction Corner Finding Track Fitting Kalman Filter Calorimetry Particle Flow
6. DBSCAN* ‣ ‣ Density based clustering ‣ ‣ Neighbourhood (radius ε) around a point must have ≥ N min points ‣ ‣ Clusters formed from density reachable points *Sander et al., Data Mining & Knowledge Discovery 2, pp169–194 (1998) N min ≥ 5
7. DBSCAN in ArgoNEUT Kinga Partyka (Yale/ArgoNEUT) colours represent clusters
8. OPTICS* ‣ ‣ Extension of DBSCAN ‣ ‣ Clusters on all scales of neighbourhood ε ‣ ‣ ε scale can be tuned to minimise overclustering Ankerst et al., ACM SIGMOD Int. Conf. on Management of Data pp49–60 (1999) DBSCAN “overclusters” near vertex
9. Cellular Automata ‣ ‣ Dynamical system with rules to update local state at discrete steps cluster cell 1 cell state 1 2 3 2 1 Rule: Increment state if neighbours consistent with straight line determined by max. angle θ θ ≤ θ max θ > θ max
10. Cellular Automata 0 20 40 60 80 100 120 140 160 180 -800 -600 -400 -200 0 -180 -160 -140 -120 -100 -80 -60 -40 -20 0 ‣ ‣ CA applied to Genie CCQE ν μ interaction tracked through Geant4 proton, muon, Michel electron, products of hadronic reinteraction of proton
11. Key Point Detection ‣ ‣ Structure tensor of energy deposits can be used to identify interest points ‣ ‣ Foestner/Noble corner detection measure proton endpoint primary vertex δ electrons along μ track B. Morgan, JINST 5(07) p7006 (2010)
12. Vertex Finding in ArgoNEUT ‣ ‣ Key point detection has been used to find vertices in ArgoNEUT events
13. Particle Flow Calorimetry Latte simulation (Ben Morgan, Warwick) + Pandora PFA (Mark Thomson & John Marshall, Cambridge) 500 MeV neutral pion Lamu modeller + PandoraPFA reconstruction
14. Do LAr-TPCs Dream of Electric Reconstruction? ‣ ‣ Reconstruction algorithms in development ‣ ‣ Key point finding critical for identifying features in ν interactions ‣ ‣ Focus reconstruction efforts towards phenomenology ‣ ‣ Warwick are contributing significantly to worldwide software efforts