Hit Primitives & Basic Clustering for Supernova Triggering - PowerPoint PPT Presentation

Hit Primitives & Basic Clustering for Supernova Triggering Alexander Booth Overview: • Triggering based on individual hits. • Triggering based on clusters in channel and time. DAQ Sim Weekly Meeting. December 18, 2017 1

Analysis Details GOAL: Understand supernova triggering e ffi ciencies and corresponding background rates for di ff erent levels of trigger. Explore the potential to trigger on just the HIT PRIMITIVES of individual hits and compare this performance to CLUSTERING hits in channel and time. • Using an amended version of the DAQSimAna (M. Baird, K. Warburton) module in dunetpc. • Running on files produced for the DUNE physics week, SN+radiologicals +noise. • Non-compressed, 1000 events each of 1 drift window and containing 1 MARLEY neutrino per event. Include Ar42. 1x6x2 geometry . /pnfs/dune/persistent/users/talion/v06_56_00/reco/snb_bkg_nocompression_dune10kt_1x2x6/ files.list • Gauss hit finder to pick out hits. All collection plane. Save hit primitives such as hit time, ADC sum of hit, hit RMS etc. Why not fast hit? (For now). • Backtrack each of these hits to a generator - was it radiological/noise/ supernova. 2

Why not fast hit? Problems with Hit Size = End time - start time • Fast hit looks for time bins with ADC above a user defined ADC threshold. >2 bins above threshold, calls it a hit. • Start time = first bin above threshold, end time = last bin above threshold. Results in many ‘skinny’ hits in time -> Many hits not correctly backtracked. Generator Type, Fast Hit (20ADC) Generator Type, Gauss Hit 3 3 10 htemp htemp 10 htemp htemp × × Number of Hits Number of Hits Entries Entries 307076 307076 Entries Entries 396728 396728 200 Mean Mean 1.409 1.409 Mean Mean 4.227 4.227 300 Std Dev Std Dev 2.083 2.083 Std Dev 0.9519 Std Dev 0.9519 180 160 250 140 200 120 100 150 80 100 60 40 50 20 0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 Generator Generator Generator type 0 is ‘noise’. 3

Gauss Hit Originally developed by Jonathan Asaadi but expanded by many. • Finds pulses in each view above individually configured thresholds. • ‘Touching’ hits on a channel are merged up to a configurable max. • Hits fit to a gaussian peak for: ‣ Start and end time. ‣ Peak time. ‣ Peak ADC. • Total hit ADC is integral of raw data, not fit by default. • Default generous max Chi^2 for allowed hits. Has a hard coded hit size minimum of 5 ticks. 4

Hit Primitives 5

Marley v Radiologicals v Noise Comparing MARLEY, Radiologicals & Noise: HitSADC Comparing MARLEY, Radiologicals & Noise: HitPeak Number of Hits Number of Hits MARLEY MARLEY 5 10 5 Radiologicals Radiologicals 10 Noise Noise 4 10 4 10 3 10 3 10 2 10 2 10 10 10 1 1 0 200 400 600 800 1000 0 20 40 60 80 100 HitSADC, (ADC) HitPeak, (ADC) Comparing MARLEY, Radiologicals & Noise: HitRMS Number of Hits MARLEY 5 10 Radiologicals Noise 4 10 Some variables show fairly 3 10 good separation. 2 10 Potential to make pre-clustering cuts. 10 1 0 1 2 3 4 5 6 7 8 HitRMS, (ticks) 6

Marley v Individual Generator Comparing MARLEY to Individual Radiologicals & Noise: HitSADC Comparing MARLEY to Individual Radiologicals & Noise: HitPeak Number of Hits Number of Hits Noise Noise MARLEY MARLEY 5 APA frame, Co60 APA frame, Co60 10 5 10 CPA fram, K40 CPA fram, K40 Ar39 Ar39 n n Kr Kr 4 10 4 10 Po Po Rn Rn Ar42 Ar42 3 10 3 10 2 10 2 10 10 10 1 1 0 200 400 600 800 1000 0 20 40 60 80 100 HitSADC, (ADC) HitPeak, (ADC) Comparing MARLEY to Individual Radiologicals & Noise: HitRMS Number of Hits Noise 5 10 MARLEY APA frame, Co60 CPA fram, K40 Ar39 n Ar42 4 10 Kr Po Rn Ar42 3 10 2 10 10 1 0 1 2 3 4 5 6 7 8 HitRMS, (ticks) 7

Triggering on hit primitives Cut: - HitRMS<1.8TDC, HitPeak >=20ADC - HitRMS>=1.8TDC, HitPeak>exp(-HitRMS+5) Define trigger: Number of SN like hits in an event. SN e ffi ciency: Did we have a trigger due to Marley hits? Background rate: Count number of triggers due to backgrounds. % of hits left in the sample after cut applied. Background Rate: HitSADC v HitRMS Cut Background Rate: HitSADC v HitRMS Cut Supernova Trigger Efficiency: HitPeak v HitRMS Cut Supernova Trigger Efficiency: HitPeak v HitRMS Cut 1 1000 Efficiency Rate, (Hz) Radilogicals: 0.15% 0.9 Marley: 69% Noise: 0.34% 0.8 800 0.7 600 0.6 0.5 400 0.4 0.3 200 0.2 0.1 0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 8 Number of Supernova Like Hits Number of Supernova Like Hits

Clustering in Channels & Time 9

Clustering Algorithm • Hits ordered sequentially by channel. Walk along the wires looking for hits on ‘adjacent’ channels. Can modify adjacent channel tolerance (ACT). Channel 7 1 4 8 9 2 3 5 6 10 If ACT = 2 X X X X X X ACT • Calculate the total ADC sum of the hits in the cluster . • Can then cut on minimum number of channels in a cluster or cluster width . • Order hits sequentially in time within each cluster. Walk through hits, checking time separation. Can modify adjacent hit time separation (ATS). • Look for a number of sequential time hits. Number of adjacent time hits. If required number of Time adjacent time hits > 2 X X X X X X X PASS ATS X X X X X X FAIL 10

Cuts/Parameters Total ADC Sum per Cluster Number of Clusters 5 10 MARLEY Backgrounds 4 10 3 10 Anything in blue is a parameter that can be changed/cut on. 2 10 10 1 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 HitSADC Total, (/Cluster) Number of Hits per Cluster Number of Channels per Cluster Number of Clusters Number of Clusters MARLEY 4 MARLEY 10 4 10 Backgrounds Backgrounds 3 10 3 10 2 2 10 10 10 10 1 1 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Number of Channels Number of Hits, (/Cluster) 11

E ffi ciency and Background Rates Single Marley Neutrino in a 2.246ms drift window. Background Rate, Varying Minimum Number of Channels With Hits in a Cluster Background Rate, Varying Minimum Number of Channels With Hits in a Cluster Supernova Trigger Efficiency, Varying Minimum Number of Channels With Hits in a Cluster Supernova Trigger Efficiency, Varying Minimum Number of Channels With Hits in a Cluster Efficiency Rate, (Hz) 1 70000 60000 0.8 50000 0.6 40000 30000 0.4 20000 0.2 10000 0 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 Minimum Channels With Hits in a Cluster (#channels) Minimum Channels With Hits in a Cluster (#channels) ROC Curve, Varying Minimum Number of Channels With Hits in a Cluster ROC Curve, Varying Minimum Number of Channels With Hits in a Cluster Pick a parameter to float and keep the SN Efficiency 1 others loose. 0.8 Define trigger: Is there a cluster which passes 0.6 all of these cuts? 0.4 SN e ffi ciency: Did we have a trigger from a cluster with > 2 Marley hits? 0.2 Background rate: Count number of triggers 0 0 10000 20000 30000 40000 50000 60000 70000 Background Rate, (Hz) due to backgrounds. 12

After tuning: After looking at each of these e ffi ciency plots by eye to select the ‘best’ value of each variable. Adjacent Minimum Number of Adjacent time Cluster ADC Background channel channels in a adjacent time E ffi ciency separation Sum Rate (Hz) tolerance cluster hits. 3 2 20 2 350 92% 18.0 3 2 50 3 500 83% 0 13

Back to the Data Find more hits per event with a di ff erent hit finder? Fair number of events below 10MeV. 14

Summary • Applying a preselection to the hits based on the hit primitives is quite powerful and greatly cuts down the number of hits that would need to go through clustering. • It is possible over 1000 events to get the background rate down to zero at 83% single Marley neutrino e ffi ciency for 5<Nu E<45MeV. • At 92% e ffi ciency background rate 18Hz for this ‘mini’ detector geometry. To high. • Need more stats and to use the fast hit finder to investigate the e ff ect of hit ADC threshold. 15