Electron Sign Determination in GENIE v e Events in Magnetized GAr Tom Junk DUNE ND Meeting January 11, 2018 The question: What fraction of electrons in v e CC events have their signs correctly identified? Approximate answer: We expect this to be very good for a GArTPC in a magnetic field. But very short tracks are hard to measure well.
Setup: Simulation • I'm using LArSoft with the 1x2x6 TPC geometry (7.2m x 6m x 15 m) too big! • Set the density of LAr to that of GAr. • Set the magnetic field to 0.4 T • Run GENIE with nominal task force flux file: gsimple_dune10kt_v1_1x2x6_g4lbne_v3r4p2_QGSP_BERT_run15_12388_80GeV_neu trino_*.root • Flux is not oscillated, so flavor and energy composition is close (modulo geometric acceptance effects) • Generate only v e and v e events (CC and NC). NC will be not representative because the v e spectrum is harder than v 𝜈 . Sample size = 2000 events. • NC background is small in this study (but I'm being naive) • GEANT4 simulation of GENIE particles • No detector response simulation – looking at MCParticle trajectory steps. Bounded by voxel size of 0.3 mm. Also containment isn't right as the 1x2x6 APA detector is too big by far. Can always apply fiducial cuts in the ND. 2 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Setup: "Reconstruction" and Analysis • Cheat the primary vertex using the MCTruth Neutrino position (it's the end of the neutrino's trajectory). • Cheat electron identification by looking at PDG codes • Pick the highest-energy electron or positron which starts within d cut of the primary vertex (whether or not it is the electron from a charged- current interaction or something else) • Make plots as function of the true electron energy • Select trajectory points that are isolated from other particles' trajectory points. To be considered for selection, a trajectory point must be at least r cut in 3D from any trajectory point on any other charged particle, including delta rays. But if a hit is within 1mm of the track, consider it indistinguishable. G4 makes lots of short delta rays with LArSoft settings. 3 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Short Delta Rays in GEANT • With LArSoft settings, GEANT4 conservatively creates lots of very short delta rays (thousands per event). Most are a few hundred microns in length. Electron Track Length (cm) Electron Track Length (cm) Linear and log scales. Track lengths are just endpoint – start point distance; Affected by the magnetic field (but not the X component). 4 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Setup: Analysis cont'd • From the list of isolated trajectory points, compute the sagitta incrementally: - Start from the beginning of the track and for each point i, calculate the maximum perpendicular distance of a track point from the line between the beginning to point i • Calculate sagitta in the ( y,z ) plane. B is along x . • If any sagitta calculated exceeds s cut in absolute value, use the signed sagitta (computed with a cross product) to determine the sign of the track. Keep stepping through points i until we run out. Provides protection against looping tracks. • The values of d cut , r cut , and s cut depend on detector parameters and hide a lot of ignorance in the procedure. 5 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Fiducial Volume Cuts • New since Jan 5. • Primary vertex must be 30 cm away from the APA plane, 30 cm away from the CPA's, and 1.5m away from the downstream field-cage wall. 30 cm from other field-cage walls. • Previous iteration had some electrons showering in the steel frames of the APA's and in the cryostat material 6 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Results Example • Initial guess: d cut = r cut = s cut = 0.5 cm. Maybe a little small, due to the 7mm pixel dimension along one axis. • Needed a log scale to show small contributions • Track length is an important determinant of whether we can measure the sign • Here, | 𝛦 x| < 2.5m, | 𝛦 yz|< 2.5 m • Very small chance to not be able to tell what the sign is • Allowing tracks to be 5m long in both dimensions makes the "missing" component go away 7 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
An example Correct-Sign Event Event display centered on primary vertex Magenta: Electrons Blue: Protons Red: Pions Material interactions Black: Muons not classified (APA's cryostat) Electron goes ROOT/X drawing backwards a bit bug puts magenta but still is measured pixels on protons. properly. 8 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
An example Incorrect-Sign Event This one fails the fiducial cuts now Event display centered on primary vertex Magenta: Electrons Blue: Protons Red: Pions Material interactions Black: Muons not classified (APA's cryostat) 9 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Shorter Tracks: 1 m long • Tried looking at only the first 1m in YZ and 1m in X of tracks. • Keep d cut = r cut = s cut = 0.5 cm • Makes a big difference! • A small detector with poor sagitta resolution struggles with the electron sign • Kind of a test to make sure the program isn't crazy 10 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
ALICE • ALICE has 5m diameter, and a space point resolution perpendicular to the B and E fields and the beam of about 1 mm (between 0.4 and 2mm). B=0.5T in ALICE. Up to 159 space points per track C. Lippmann, Physics Procedia 37 (2012) 434-441 11 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Allow Smaller Sagittas • YZ length < 1m and X length < 1m • d cut = r cut = 0.5 cm, s cut = 0.1 cm 12 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Even Shorter Tracks • YZ length < 50 cm and X length < 50 cm • d cut = r cut = 0.5 cm, s cut = 0.1 cm 13 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Sagitta cut down to 100 microns • Still with a YZ track length cut of 50 cm and X track length of 50 cm • ALICE gets up to 159 points to fit. Averaging is strong. • But ALICE tracks are 2.5 m long • Our primary vertices are distributed through the detector. A 50 cm track will not have lots of hits for averaging. 14 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Not done yet, but summary of conclusions • Biggest effect is from making sure that an electron has a measurable bend in the detector volume - Strong enough field – what is the minimum requirement? - Long enough track - Space-point resolution • Multiple scattering seems not to be a problem - otherwise we'd have more incorrect-sign assignments at low sagittas • Neutral Current BG seems not to troublesome - but I cheated the primary vertex location - no pileup. - would need to topologically identify conversions which are displaced from NC events but which might look like their own events or pileup. Two electrons, plus pointing to a hadronic recoil. 15 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Two NC events with electrons Distances are marked off in cm 16 Jan 11, 2018 Tom Junk | GArTPC Electron Sign ID
Recommend
More recommend
