SBND (UK)* software and physics report Dom Brailsford on behalf of SBND UK DUNE UK meeting at Manchester 06/03/19 Covers topics not already presented at this meeting*
The Short Baseline Near Detector (SBND) • Serves as near detector of the • LArTPC short baseline neutrino program • 112 t fiducial volume • 110 m from BNB target • Sterile neutrino search � 2
Motivation p Neutrino vertex Time, TDC p p SBND ν μ CC 0π 3p signal μ - event in the SBND MC sample GENIE v02.12.10, The Short Baseline Near Detector (SBND) Default+MEC Wire number • SBND will observe millions of R. Jones neutrino interactions • Measurements of many exclusive cross section channels ν μ CC exclusive Proton multiplicity ν μ CC inclusive event event rate rate breakdown, in the true ν μ CC • BSM physics not stacked breakdown, final state stacked R. Jones Neutrino energy reconstruction ● SBND will observe huge statistics Low momentum particles ○ at bubble-chamber resolutions (3 mm). High multiplicities ○ Neutrino interaction studies ● Ensuring the software is capable of detecting 2p-2h cross-section measurement ○ particles in such detail is crucial for Detecting and analysing rare final states ● � 3
SBND physics organisation • Geared towards SBND’s own physics program i.e. what can SBND measure on its own • Cross sections • BSM physics • Detector measurements • Organised via the Physics and Analysis Tools (PAT) group • 2/4 convenors are UK-based • Andrzej Szelc • Costas Andreopoulos • Heavy UK involvement � 4
SBN physics organisation • Shift to more formal SBN group structure • 10 groups 10 sub-groups • 1 SBND and 1 ICARUS convener per group • Aim: Unify sim./reco./ analysis for the multi- detector physics program � 5
SBN physics organisation UK leadership 10 sub-groups Strong UK contribution � 6
D. Brailsford MC production • 3 production campaigns runs between 2018-2019 • We took a production processing hiatus in mid-2018 to switch from project.py-based productions to POMS-based productions • Manual submission -> automatic submission • Manual job recovery -> automatic recovery • Complete handling/bookkeeping of Monte Carlo file metadata • Automatic transfer of files to tape for permanent storage • Processing on the open science grid as well as FNAL • SBND is currently running its largest ever production campaign • ~1,000,000 events for physics analysis � 7
D. Brailsford SBND-POMS production workflow Output files FTS Initial jobs POMS copied submitted box CPU Jobs report finished running Files copied using SAM File metadata bookkeeping fife-wrap declared SAM Tape Jobs auto-submitted using SAM datasets from tape � 8
A. Ezeribe BNB timing structure in LArSoft Time structure of our current The actual time structure of neutrino simulation the BNB • First implementation feature complete • Modelling the BNB timing structure in LArSoft would allow us to study: • Timing structure correctly propagates through the entire simulation change • Detector timing resolution • Now finalising before making available to • BSM physics the community � 9
T. Brooks Cosmic Ray Tagger (CRT) system reconstruction Cosmic ray tagger system reconstruction CRT reconstruction The SBND TPC is almost completely covered by a number of cosmic ray taggers formed of plastic scintillator. They can be used to reconstruct CRT hits (the position and time of particle-CRT intersections) and CRT tracks (the trajectory of through going particles). CRT-TPC matching TPC reconstructed tracks can be They can be matched to CRT matched to CRT hits by projecting tracks by comparing angles their ends on to the CRT taggers. and start/end positions. � 10
T. Brooks CRT cosmic background removal CRT cosmic background removal Background removal Fiducial volume : Remove tracks with start and end points ● within 10 cm of TPC walls. TPC topology cosmic ID : Tracks reconstructed outside of TPC. ● Preliminary results Tracks which start outside of fiducial volume ● and stop inside it. Sample of 5,000 TPC contained neutrino Tracks which match CRT tracks. ● events with a corsika cosmic overlay. t0 tagging cosmic ID : Only using CRT matching and basic TPC information (no light or Pandora t0 from stitching tracks across the CPA. ● reconstruction). t0 from tracks which match CRT hits. ● t0 from matching tracks with APA crossing ● Able to go from a 1:14 neutrino muon to points. cosmic muon ratio to 2:1. � 11
D. Barker Shower reconstruction validation • SBND is currently investigating various combinatorics of reconstruction algorithms • Blurred cluster (M. Wallbank) • EMShower (M. Wallbank) • Pandora (Team pandora) • SBND has developed a shower validation module • Easily compare di ff erent emshowerNew : Pandora + EMShower reconstruction algorithms pandoraShower : Full pandora reco. emshowerBLUR : Blur. cluster + EMShower • Easily benchmark reconstruction performance using metrics � 12
D. Barker Shower reconstruction validation Example electron reconstruction emshowerNew : Pandora + EMShower pandoraShower : Full pandora reco. emshowerBLUR : Blur. cluster + EMShower � 13
Reconstruction E. Tyley tuning • Aspects of pandora are tune-able by end users • Track-shower separation • Vertex identification • Currently running a phase of exploratory track-shower Before After separation tunings • Focussed on shower segmentation for electron particle gun • We are now ramping up a metric-based tuning approach with a more realistic topology � 14
E. Tyley Recombination • Aim is to understand how recombination a ff ects calorimetric reconstruction in showers • Truth-based study using particle gun electrons and muons • A flat recombination factor for electrons is appropriate, which di ff ers to muons • Study extended to hit-based calorimetry using the same samples • Sums reco. energy for every hit • Reconstructed energy as a function of true electron momentum shows a modest dependence � 15
J. Tena-Vidal PID via track follow-down Muon vs pion ID in LArTPC detectors • Calorimetry-based PID struggles to Need to define a PID method for muons and pions distinguish similar mass particles such as based on a track follow down procedure muons and pions Calorimetry PID methods are not good to ● • Other options are needed to separate distinguish particles with similar mass muons and pions However, muons and pions leave different ● signatures in the TPC • ~70% of muons are captured before 73% of the times the muon is absorbed - decay - simple straight line track ○ topology straight line track Michel electrons can be used to tag muon ○ • Decaying muons can be tagged by the tracks Michel electron Pions may interact with the environment by ○ absorption, elastic or inelastic scattering • Pions are more inclined to scatter in The final goal is to apply this to 𝜉 𝜈 CC1π + events to the TPC ● improve the analysis • Goal is to develop a pion/muon separator to improve the CC ν μ 1 π selection 1 J.Tena Vidal - University of Liverpool � 16
J. Tena-Vidal PID via track follow-down • Method initially developed by MicroBooNE for Michel electron searches • Pandora provides a parent/child hierarchy • Based on calculating Pearson coe ffi cient for hits within a search window � 17
R. Jones CC ν μ selection μ Topological selection Use calorimetric information to ● distinguish protons from π & μ Along with geometrical ● Single interaction ν μ BNB-only information for MIPs n Fiducial volume of the TPC ↓ True / Reco → CC Inc. CC 0π Initial priority on ● p π μ tagging CC 0π 39,100 32,650 π Small amount of ● p Selection CC 1π focus on protons 8,386 3,218 n ν optimised for μ μ & proton purity CC Other 658 70 Total BNB-only events with a single NC 2,967 2,130 63,830 contained, reconstructed neutrino vertex Efficiency 92.0% 76.9% True vertex also contained 96.3% Purity 94.2% 85.8% Maximum 1 escaping track 99.9% Exactly 1 escaping track 5.5% Efficiency: Signal topology / Total true topology Of these, only the true muon escapes 95.9% Purity: Signal topology / Total selected topology 2 5.3% of events are basically free, guaranteed muons! � 18
Recommend
More recommend
