Pandora pattern recognition p for LArTPCs L. Escudero for - PowerPoint PPT Presentation

Pandora pattern recognition γ γ p for LArTPCs µ L. Escudero for the Pandora Team & the MicroBooNE and DUNE collaborations Calibration and Reconstruction for LAr TPC Detectors Thanks to John Marshall, Andy Blake and Steve Green for contributing (voluntarily or involuntarily) to these slides 1

Pandora Pattern Recognition A single clustering approach is unlikely to work for complex topologies (mix of track and shower-like clusters) Instead, the Pandora project is a novel John Marshall method of pattern recognition, which tackles this project from its beginning in ILC and LHC using an advanced multi- algorithm approach: • Build up events gradually Mark Thomson Andy Blake • Each step is incremental - aim not to make mistakes (hard to undo) • Deploy more sophisticated algorithms as picture develops • Build physics and detector knowledge into algorithms • Possible thanks to the Pandora Software Development Kit for Pattern Recognition (Eur. Phys. J. C 2015, 75: 439) for all use cases 2 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pattern Recognition Low level High level Reconstruction Reconstruction Reconstruction Noise Filtering Track Fitting path to Physics Signal Processing Calorimetry Hit Reconstruction Particle ID This is us! Pattern Recognition Clustering 2D -> 3D Particle Hierarchy Our output : hierarchy of reconstructed particles Our input : collection of 2D hits 3 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pattern Recognition Building up events gradually, with chains of small algorithms, harnessing a number of powerful capabilities: • 120+ algorithms and tools This is us! • Use of multiple parallel hypothesis Pattern Recognition • Iterative reconstruction Clustering techniques 2D -> 3D • Allowing incorporation Particle Hierarchy of ML/DL methods to make algorithm decisions 4 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora LAr TPC Pattern Recognition • Our aim is to provide automated pattern recognition for general usage (any particle, any topology) • Pandora algorithms are reusable for different single- phase LArTPCs • Focus in recent years mainly on MicroBooNE and now protoDUNE, but expanded to also SBND, ICARUS, DUNE FD LArTPC reconstruction: high quality, complex images, and lengthy drift times (i.e. long exposures) meaning a significant cosmic-ray background for surface detectors. 5 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora Consolidated Output Two di ff erent chains with tens of algorithms each: • Cosmic ray reconstruction (Pandora Cosmic) • Neutrino interaction/test beam reconstruction (Pandora Neutrino) Harnessing the chains of algorithms in an intelligent manner to Andy Smith provide a consolidated output: Input Output *using SVM/BDT trained models 6 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora LAr TPC Pattern Recognition A little bit deeper into some of the steps in the algorithm chains… 2D track reconstruction a) We start by producing a list of 2D clusters (per plane) that represent continuos, unambiguous lines of hits, starting/stopping at each branch or ambiguity. b) Then series of cluster-merging and cluster-splitting algorithms evaluate the list of 2D clusters and change them based on topological information, carefully aiming at safe merges, improving completeness without compromising purity. Example of cluster-merging algorithm “in action” Can’t do justice in a few slides, please find more in the Pandora MicroBooNE paper (Eur. Phys. J. C 78, p82 2018) 7 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora LAr TPC Pattern Recognition A little bit deeper into some of the steps in the algorithm chains… 3D track reconstruction Cluster-matching ambiguities are 2D clusters in the three planes are compared identified by tools and used to to find those representing same particle, “diagonalise” the tensor Tools modify exploiting the common drift-time coordinate 2D clusters as appropriate and then run and our understanding of wire plane again from the beginning on the geometry. Results are stored for each 3x2D updated tensor. Example tool: combination of clusters in a rank-three tensor. The Pandora Rotational Coordinate Transformation System is a very powerful feature, allowing 2D->3D and 3D->2D projections 8 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora LAr TPC Pattern Recognition A little bit deeper into some of the steps in the algorithm chains… Neutrino Interaction Vertex A key algorithm is the one to select the most appropriate 3D vertex position from a list of candidate vertices. Used first a simple score, then a more sophisticated one (with topological and charge asymmetry information) for each candidate. Now a multivariate approach (SVMs) is used for MicroBooNE. This is an example of how powerful the multi-algorithm approach is, by breaking down pattern recognition into small problems, even allowing to use Machine and Deep Learning methods to solve some of them! Downstream usage: • Split 2D clusters at projected vertex position, and use vertex to protect primary particles This allows us to have good performance in Jack Anthony interactions with many final state particles… Raquel Castillo NuInt talk 9 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora LAr TPC Pattern Recognition A little bit deeper into some of the steps in the algorithm chains… Shower Reconstruction 3D - Reuse ideas from the track tensor, using envelopes and their projection to 2D - A key step in the pattern recognition is to match 2D shower clusters characterise 2D clusters as track/shower like using topological information, to identify shower spines and allow to grow branches (nearby shower-like clusters), whereas prevent doing so around track-like clusters. 10 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora LAr TPC Pattern Recognition 3D hit/cluster reconstruction Iteratively, produce smooth trajectory Y Y For each 2D hit, sample clusters in other views at the same X, find u in , v in , w in and use analytic expression to find the most consistent 3D space point by X X minimising a χ 2 function. Z Z 2D/3D Particle refinement and 3D particle hierarchy Several algorithms deal with remnants to improve Finally, walking particle completeness, (esp. sparse showers). Sliding backwards linear fits are used to define 2D envelopes and 3D from interaction cones for picking up small clusters/fragments. vertex, use 3D clusters to organise particles into hierarchies (building parent- daughter links) 11 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora Performance Metrics E ffi ciency: Correct event: Fraction of target MCParticles Really strict metrics, events only correct if ( reconstructable* ) with at least each MC target ( reconstructable* ) particle is one matched reconstructed matched to exactly one reconstructed particle (fulfilling conditions based on matched hits) particle, with correct parent-daughter links. Event1: correct Event2: incorrect * see more precise definitions in paper *labels correspond to true MC particle type 12 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora Performance in MicroBooNE The Pandora multi-algorithm approach to automated pattern recognition of cosmic-ray muon and neutrino events in the MicroBooNE detector (Eur. Phys. J. C 78, p82 2018) CC RES: μ +p+ π + Performance on selection of Exclusive Final States “Correct” here really strict, means ~perfect event! 13 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Pandora in Analysis & Detector Physics Pandora reconstruction is used in multiple Comparison of ν μ − Ar multiplicity distributions MicroBooNE analyses, and also in calibration observed by MicroBooNE and detector physics studies (calorimetry, to GENIE model SCE, lifetime, di ff usion…) predictions (paper) Aleena Rafique Detector calibration using through going and stopping muons in the MicroBooNE LArTPC (public note) Multiplicity 2, w/o particle ID Colton Hill Automated Selection of ν e from NuMI These are just some examples, see more in using MicroBooE MicroBooNE public docs web (NuInt talk) 14 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Exciting times for protoDUNE! Pandora reconstruction on protoDUNE real data! Steven Green Very e ffi cient reconstruction in terms of memory and CPU time (tipically <1 minute for protoDUNE data events) 15 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors

Exciting times for protoDUNE! • Using beam trigger information (trigger recob::Track providing position and direction) performance on real data can be explored • E ffi ciency: Fraction of target events where the trigger is active and we reconstruct at least one test beam particle • This folds in: cosmic vs test beam separation, reconstruction, tagging • Main reason for ine ffi ciency is cosmic contamination, resulting in test beam hierarchy reconstructed as cosmic ray protoDUNE SP DATA protoDUNE SP simulation Note: Total number of hits (U+V+W) 16 Lorena Escudero, Calibration and Reconstruction for LArTPCs Detectors