Supernova Burst Trigger Studies in DUNE FD Single Phase TPC Alexander Booth Collaboration Monthly Meeting. April 20, 2018 1
Overview • Supernova burst time and energy profiles. • Finding individual neutrino interactions - clustering algorithm. • Performance of a simple SN burst trigger - galactic coverage and fake rate. • DAQ requirements - time to record. 2
Strategy & Assumptions 3 Fast ‘back of the envelope’ approach to establish broad features of DUNE’s ability and DAQ requirements to capture neutrinos from a SN. Many questions and assumptions still to be addressed. DESIGNED TO BE MODULAR: Background and Hit finder/ Burst trigger Astronomical SN models. noise models. hit clustering. design. models. • 1 specific supernova • 8 radiological backgrounds. • Number of events expected. model used. -Are there more? • Distribution of SN candidates - Hudepohl model. -Correct rates? in the galactic neighbourhood. - 11.2 solar mass • White noise. -Simple minded addition of progenitor. -Coherent noise? LMC to the distribution. • E ff ect of oscillations? • Not considered cosmics.
Distribution of Supernovae 4 SN Probability Distribution Mirizzi, Raffelt & Serpico, astro-ph/0604300 0.06 Milky Way 0.05 Large Magellanic 0.04 Cloud 0.03 0.02 Empty space until 20 events Andromeda! (780 kpcs) 0.01 0 10000 events 800 events 0 10 20 30 40 50 Distance, (kpc) Define ‘galactic neighbourhood’ as Milky Way + LMC We can reasonably consider issuing a burst trigger for SN in this region.
Supernova Event Generator - MARLEY 5 Marley Time Profile, Event Normalised. First 50ms Marley Time Profile, Event Normalised. First 50ms 3 − 10 × 0.2 Events Events 0.09 Oscillations will suppress 0.18 the number of events in the 0.08 0.16 first 10 ms. 0.07 0.14 0.06 Events out to ~10 s in 0.12 current simulation. 0.05 0.1 0.04 0.08 0.03 0.06 0.02 0.04 0.01 0.02 0 0 0 5 10 15 20 25 30 35 40 45 50 0 0.01 0.02 0.03 0.04 0.05 Time, (s) Energy, (MeV) Only nue-CC on Ar nucleus, Time, (tick) 220 7 individual 260 (~75% of total). events drawn on 200 240 1 event display 180 220 160 True primary lepton energy peaks 200 140 ~ 10 MeV. 180 120 100 160 80 1 channel ~ 0.5 cm, 1 tick ~ 0.08 cm 140 60 120 40 ‘Typical event’: 5 cm x 5 cm 100 20 80 0 9440 9450 9460 9470 9480 9490 9500 Charge Channel, (Channel No)
Radiological Backgrounds 6 Simulation contains white noise and radiological backgrounds. In current simulations, backgrounds are dominated by radiologicals, not noise. Require a better understanding of these backgrounds. Source Notes Ar39 Intrinsic to LAr Ar42 Intrinsic to LAr Co60 APA frame Ur-238 in concrete Neutrons Po Simulates Rn daughters on PDs K40 CPA frame Kr85 LAr Rn222 LAr (contamination) Dominant background of burst trigger. https://www.overleaf.com/13924050chkrxfmktthr#/53974837/
Clustering Individual SN Neutrinos 7
Clustering Algorithm 8 Takes channel ordered hits from a hit finder (currently Gauss hit). CLUSTER IN CHANNEL AND TIME SPACE • Hits ordered sequentially by channel. Walk along the wires looking for hits on adjacent channels. • Within these channel cluster, group hits close in time. • Cut on total ADC sum of the hits in the cluster, minimum number of channels in a cluster or cluster width . • Finally require a certain number of hits in a cluster. Channel 1 7 2 3 4 8 9 10 5 6 Require > 3 1 X X X 2 hits. Time X X 3 4 X X CLUSTER 5 6 X NO CLUSTER 7 X 8 X 9 It’s fast: Given the assumptions e.g. an ordered list of hits was provided, the clustering for 10kt could be run on a single CPU. https://indico.fnal.gov/event/16859/contribution/1/material/slides/0.pdf
E ffi ciency & Background Rates 9 Efficiency 1 Smaller clusters More background 0.8 Larger clusters Less 0.6 background Efficiency & 10kt Background Rate Eff: 0.91, Bkgd: 19.11Hz 0.4 Eff: 0.88, Bkgd: 5.81Hz Eff: 0.86, Bkgd: 3.87Hz Eff: 0.81, Bkgd: 1.66Hz 0.2 Eff: 0.70, Bkgd: 0.43Hz Eff: 0.58, Bkgd: 0.10Hz 0 0 5 10 15 20 25 30 35 40 45 50 True Neutrino Energy, (MeV) Input to burst trigger: Di ff erent clustering algorithms allow the trade o ff between lower e ffi ciency and background rate to be explored.
Supernova Burst Trigger Clustering e ffi ciency & SN / radiological / Burst background acceptance Hit-clustering trigger noise simulations di ff erent clustering configurations 10
SN Burst Trigger 11 Unique signatures of SN burst: • Events spread out over a long time (exponential cooling of SN with 2-3 second decay time). • Events typically higher energy than background. Strategy: keep it very simple for now. • Count the number of hit-clusters in a 10 second window. • Trigger above a threshold number of hit-clusters. Fake triggers: • Use background rate from clustering algorithm, assume it fluctuates in Gaussian way. • Can map out burst-trigger rate as a function of threshold number of hit- clusters.
Result: Galactic Neighbourhood Coverage 12 Assuming a fake burst trigger rate of 1/month, what is our SN sensitivity vs. distance? Galactic Neighbourhood Coverage, Fake Trigger Rate 1/Month Galactic Neighbourhood Coverage, Fake Trigger Rate 1/Month Efficiency x SN Probability 1 − 10 Milky Way 2 − 10 Galaxy edge, harder Close SN, easy to trigger 3 − to trigger 10 4 − LMC 10 − 5 10 Require larger 6 Individual Marley Efficiency & 10kt Background Rate − 10 clusters to Eff: 0.91, Bkgd: 19.11Hz Eff: 0.88, Bkgd: 5.81Hz increase Eff: 0.86, Bkgd: 3.87Hz Eff: 0.81, Bkgd: 1.66Hz sensitivity 7 − 10 Eff: 0.70, Bkgd: 0.43Hz Eff: 0.58, Bkgd: 0.10Hz SN Probability 8 − 10 0 10 20 30 40 50 SN Distance, (kpc) Conclusion: can trigger on nearby SN easily. However, capturing the 1/5 of SN coming from LMC requires more work and is more dependent on our assumptions (e.g. hard to model neutron bkg).
Result: Fake Trigger Rate vs. Galactic 13 Previous slide required 1/month fake trigger rate. This slide shows trade o ff between e ffi ciency and this rate. 10 Fake Trigger Rate, (Hz) Individual Marley Efficiency & 10kt Background Rate 1 Eff: 0.91, Bkgd: 19.11Hz Eff: 0.88, Bkgd: 5.81Hz 1 − 10 Eff: 0.86, Bkgd: 3.87Hz Eff: 0.81, Bkgd: 1.66Hz 2 − 10 Eff: 0.70, Bkgd: 0.43Hz Eff: 0.58, Bkgd: 0.10Hz 3 − 10 4 − 10 − 5 10 1/Month 6 − 10 7 − 10 8 − 10 − 9 10 10 − 10 11 − 10 12 − 10 13 − 10 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 Galactic Neighbourhood Coverage Trigger on 98% of SN in the neighbourhood, issuing 1 fake trigger per month. In 10 kT
Time Profile Studies 14
Time Profile Studies 15 Time Since First Neutrino Passed Through Detector, (s) High data rate: 1.5 TB / s / 10kt module Closer SN, longer time. 10 Extract maximal information. < 1.00 Events Remaining 1 < 0.50 Events Remaining Further SN, < 0.20 Events Remaining 1 Coverage less time. < 1.00 Events Remaining < 0.05 Events Remaining < 0.50 Events Remaining 2 1 10 10 0.8 Supernova Distance, (kpc) < 0.20 Events Remaining Record all < 0.05 Events Remaining but 1 event 0.6 in ~ 28 s, any GNSN. 0.4 0.2 0 0 5 10 15 20 25 30 35 Recording Length, (s) Establish DAQ requirements: Non-volatile bu ff er (read-out time). Time profile studies potentially very sensitive to SN model.
Summary 16 Applied a fast ‘back of the envelope’ approach to establish broad features of DUNE’s ability and DAQ requirements to capture neutrinos from a SN. BURST TRIGGER: • Reasonable to consider issuing a burst trigger for SN in the region of the Milky Way and LMC . • In simulation shown that radiologicals are the dominant background . • Demonstrated a simple burst trigger capable of catching 98% of supernovae in the galactic neighbourhood , issuing on average 1 fake trigger 1/month . TIME PROFILE STUDIES • Shape of profile is model dependent , e.g. with/without oscillations. • Studies influence DAQ requirements. • Record all but 1 event of any galactic neighbourhood SN in ~28s . Many questions and assumptions still to be ironed out.
Backup Slides 17
Analysis Details 18 GOAL: Understand supernova triggering e ffi ciencies and corresponding background rates for di ff erent levels of trigger. • Using an amended version of the DAQSimAna (M. Baird, K. Warburton) module in dunetpc (DAQSimAna/SNAnaClustering/SNAna_module.cc). • Running on files produced at Christmas, SN+radiologicals+white noise. • Non-compressed, ~ 750000 events each of 1 drift window and containing 1 MARLEY neutrino per event. Include Ar42. 1x6x2 geometry . /pnfs/dune/scratch/dunepro/MCC10-Production/SuperNovaSamples/v06_60_00/reco/ snb_timedep_radio_dune10kt_1x2x6 • Gauss hit finder to pick out hits. All collection plane. Save hit primitives such as hit time, ADC sum of hit, hit RMS etc. • Backtrack each of these hits to a generator - was it radiological/noise/ supernova. Cluster hits in channel and time space, make geometric cuts, trigger on number of hits in the cluster.
Recommend
More recommend
