FERMILAB-SLIDES-18-085-PPD Mu2e Calorimeter Clustering Studies Emm a Castiglia with Giani Pezzullo and Sarah Demers Yale University New P erspectives 2 018 This document was prepared by [Mu2e Collaboration] using the resources of the Fermi National Accelerator Laboratory (Fermilab), a U.S. Department of Energy, Office of Science, HEP User Facility. Fermilab is managed by Fermi Research Alliance, LLC (FRA), acting under Contract No. DE-AC02-07CH11359.
Mu2e Calorimeter Overview • Signal: 105 MeV Conversio n Tracker Production Target Stopping Target Electron ( CE) w ithout n eutrinos • Requirements of Calorimeter: • Energy Resolution: ~10% • Timing Resolution : ~1ns Transport Solenoid Calorimeter • Position Resolution: 1cm • Calorimeter Role: • Particle Identification – reject bkg • Cosmic Ray Muon • Needs t o b e a t rigger f or d ata s torage! 2 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Mu2e Calorimeter Specifications • Two annular disks of undoped CsI Crystals • Dimensions: 3.4x3.4x20 cm 3 • Radiation length of 2.1 cm (~10 lengths in crystal) • Total of 1356 c rystals • Crystals are read out by 2x3 of 6x6 mm 2 UV-extended SiPM • Sign al digitized at 200MHz • Measured performance at test beam (100 MeV) • 100ps i s ti me r esolution • 5% i s e nergy r esolution 3 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Mu2e Calorimeter Clustering How it works? • Sort cryst al hits by energy deposite d and the n gro up adjacent ones • 2 a lgorithms th a t d o th is d ifferently • Full ( proto): Slower b ut m ore A ccurate – use i n o ffline r econstruction • Fast: Quicker a nd S imple– could b e t rigger i f r uns i n r eal t ime • If able to identify potential signal events , can trigger on those events 4 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Fast Algorithm for Clustering Steps: 1. Take highest energy crystal – see d w it h >10Me V 2. Look a t neighboring crystals ove r energy threshold and group - green 3. Remove crystals that are clustered together from list of crystals 4. Start process over with next highest energy crystal Modification: Fast Include next , next to next neighbors , etc Adds t ime: 1 r ing – 6 c rystals 2 r ing s – 12 c rystals 3 r ing s – 18 c rystal s 5 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Full Algorithm for Clustering Steps: 1. Take highest energy crystal – > se e d with >10MeV 2. Look a t neighboring crystals ove r energy threshold and group 3. Look at non-adjacent crystals with large energy deposits - blue -> could be deposited by photons emitted (within speed of light) 3a. Sometimes end up grouping two smaller clusters into one larger cluster 4. Remove crystals that are clustered together from list of crystals 5. Start process over with next highest energy crystal Full 6 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Goals for Clustering Studies GOAL 1 Compare Fast a nd Full Algorithms for accuracy a nd timing performance GOAL 2 Fi nd discriminating variables for improving CE selectio n from background 7 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
How do we know if we are correctly identifying conversion electrons (CE)? Use informatio n from the Virtual Detector • G4 sensitive detector right before the calorimeter that stores th e information of particles passing through i t withou t affecting them • Can get energy , radial position , etc. about the incoming CE • Truth – has c onversio n e lectro n e vents t hat ma y mi ss t he calorimeter 8 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Virtual Detector – Acceptance Acceptance: 95% Leakage: Shar p E dge - detector i s 2 0c m de e p and 3 0c m w ide ->show er loss du e to depth/reflection 3 10 × Events Energy resolution: 200 Energy of Clusters Virtual Detector 180 Fast Full : 6.4% Full 160 Fast: 8.6% 140 120 FWHM: 100 Full : 6. 2 MeV 80 60 Fast: 8. 2 MeV 40 20 0 0 20 40 60 80 100 120 9 E. Castiglia (Yale University) MeV 6/19/ 18 N e w P erspectives
Timing Comparison Hav e a few ms per e vent w ithi n w hi ch t o t rigger on o r r eject Calo Cluster Fast: ~.2ms per even t • Includes fast processing o f digitize d hits • Voltage Full Clustering: Algorithm takes 2.3ms per event • Needs Template Fit : 25ms p er event • time Complete time for each event is >27ms • => Fast Algorith m is more robust – could use as trigger 10 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Energy versus Cluster Size - Fast Distribution of Cluster Energy versus Cluster Size Fast 140 MeV 3500 CE Only: 120 3000 Peak at 80-100 Me V and 6- 9 c rystals 100 2500 80 2000 EnergyCompFast EnergyCompFast Calo Disk Entries Entries 416937 416937 60 1500 Mean x Mean x 5.757 5.757 Mean y Mean y 75.79 75.79 Std Dev x Std Dev x 2.634 2.634 40 Std Dev y Std Dev y 26.02 26.02 1000 Integral Integral 4.169e+05 4.169e+05 Skewness x Skewness x 0.6288 0.6288 Skewness y Skewness y 1.282 1.282 20 − − 500 0 0 0 0 0 0 0 0 416937 416937 0 0 0 0 0 0 0 0 0 0 inner edge Calo Disk 0 2 4 6 8 10 12 14 16 18 20 # of Crystals Distribution of Cluster Energy versus Cluster Size Fast 3 10 × 140 MeV 1200 120 Background Only: 1000 100 Peak at 10 Me V and 1 crystal 800 80 EnergyCompFast EnergyCompFast Entries Entries 2.33091e+07 2.33091e+07 600 60 Mean x Mean x 2.276 2.276 Mean y Mean y 24.46 24.46 Std Dev x Std Dev x 1.443 1.443 400 40 Std Dev y Std Dev y 11.94 11.94 Integral Integral 2.331e+07 2.331e+07 Skewness x Skewness x 2.49 2.49 Skewness y Skewness y 1 1 20 200 0 0 114 114 1 1 0 0 2.330898e+07 2 2.330898e+07 2 0 0 0 0 0 0 0 0 0 2 4 6 8 10 12 14 16 18 20 # of Crystals 11 6/19/ 18 N e w P erspectives E. Castiglia (Yal e University)
Comparing CE and bkg only - Fast ENERGY RADIAL POSITION Drop off at ~50 MeV Drop off at 410mm Distinctive background shape Normalized Events Normalized Events Radial Position of Clusters Energy of Clusters 0.16 0.06 CE only CE only bkg only 0.14 bkg only 0.05 0.12 0.04 0.1 0.08 0.03 0.06 0.02 0.04 0.01 0.02 0 0 0 20 40 60 80 100 120 140 350 400 450 500 550 600 650 700 MeV mm 12 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Comparing CE and bkg only - Fast RATIO OF SEED TO CLUSTER ENERGY Background has more clusters with only 1 crystal 0.4 Normalized Events Energy Ratio of Clusters 0.35 CE only bkg only 0.3 0.25 0.2 0.15 0.1 0.05 0 0 0.2 0.4 0.6 0.8 1 Seed Energy/Cluster Energy 13 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Conclusion and Next Steps Results so far: • Fast is quick and performs we ll enough to be used durin g triggering • Energ y r esolution of Full could be improved in Offline • Variables that could be used to differentiate: Energy and Size • Radial D istance • Ratio o f S ee d E nergy t o C luster E nergy • Next Steps: • Look at adding more rings of neighbors to Fast • Change mini mum energy cutoff for clusters – currently 10 MeV 14 E. Castiglia (Yale University) 6/19/ 18 N e w P erspectives
Questions? Thanks for listening! 15 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Backup Slides 16 6/19/ 18 N e w P erspectives E. Castiglia (Yale University)
Recommend
More recommend
