Supernova Burst Trigger Studies in DUNE FD Single Phase TPC - - PowerPoint PPT Presentation

▶
supernova burst trigger studies in dune fd single phase
SMART_READER_LITE
LIVE PREVIEW

Supernova Burst Trigger Studies in DUNE FD Single Phase TPC - - PowerPoint PPT Presentation

Nov 21, 2023 481 likes •729 views

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

slide-1
SLIDE 1

Supernova Burst Trigger Studies in DUNE FD Single Phase TPC

Alexander Booth

1

Collaboration Monthly Meeting. April 20, 2018

slide-2
SLIDE 2

Overview

2

  • 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.
slide-3
SLIDE 3

Strategy & Assumptions

3

Hit finder/ hit clustering. Burst trigger design. Background and noise models. SN models. Astronomical models.

  • 8 radiological backgrounds.
  • Are there more?
  • Correct rates?
  • White noise.
  • Coherent noise?
  • Not considered cosmics.
  • 1 specific supernova

model used.

  • Hudepohl model.
  • 11.2 solar mass

progenitor.

  • Effect of oscillations?
  • Number of events expected.
  • Distribution of SN candidates

in the galactic neighbourhood.

  • Simple minded addition of

LMC to the distribution.

DESIGNED TO BE MODULAR:

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.

slide-4
SLIDE 4

Distribution of Supernovae

4

Distance, (kpc) 10 20 30 40 50 SN Probability Distribution 0.01 0.02 0.03 0.04 0.05 0.06

Mirizzi, Raffelt & Serpico, astro-ph/0604300

Large Magellanic Cloud Milky Way

Empty space until Andromeda! (780 kpcs)

10000 events 800 events 20 events

Define ‘galactic neighbourhood’ as Milky Way + LMC We can reasonably consider issuing a burst trigger for SN in this region.

slide-5
SLIDE 5

Supernova Event Generator - MARLEY

5

Energy, (MeV) 5 10 15 20 25 30 35 40 45 50 Events 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

Time, (s) 0.01 0.02 0.03 0.04 0.05 Events 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2

3 −

10 ×

Marley Time Profile, Event Normalised. First 50ms Marley Time Profile, Event Normalised. First 50ms

Channel, (Channel No) 9440 9450 9460 9470 9480 9490 9500 Time, (tick) 80 100 120 140 160 180 200 220 240 260 20 40 60 80 100 120 140 160 180 200 220

Oscillations will suppress the number of events in the first 10 ms. Events out to ~10 s in current simulation.

1 channel ~ 0.5 cm, 1 tick ~ 0.08 cm ‘Typical event’: 5 cm x 5 cm Charge True primary lepton energy peaks ~ 10 MeV. Only nue-CC on Ar nucleus, (~75% of total).

7 individual events drawn on 1 event display

slide-6
SLIDE 6

Radiological Backgrounds

6

Simulation contains white noise and radiological backgrounds.

Source Notes

Ar39 Intrinsic to LAr Ar42 Intrinsic to LAr Co60 APA frame Neutrons Ur-238 in concrete Po Simulates Rn daughters on PDs K40 CPA frame Kr85 LAr Rn222 LAr (contamination)

Dominant background

  • f burst trigger.

https://www.overleaf.com/13924050chkrxfmktthr#/53974837/

Require a better understanding of these backgrounds.

In current simulations, backgrounds are dominated by radiologicals, not noise.

slide-7
SLIDE 7

Clustering Individual SN Neutrinos

7

slide-8
SLIDE 8

Clustering Algorithm

  • Hits ordered sequentially by channel. Walk along the wires looking for hits on adjacent

channels.

Channel

X X X X

1 2 3 4 5 6 7 8 9 10

Time

Require > 3 hits.

CLUSTER NO CLUSTER

  • Cut on total ADC sum of the hits in the cluster, minimum number of channels in a

cluster or cluster width.

8

CLUSTER IN CHANNEL AND TIME SPACE

X X X X

1 2 3 4 5 6

  • Within these channel cluster, group hits close in time.
  • Finally require a certain number of hits in a cluster.

Takes channel ordered hits from a hit finder (currently Gauss hit).

7 8 9

X X

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

slide-9
SLIDE 9

True Neutrino Energy, (MeV) 5 10 15 20 25 30 35 40 45 50 Efficiency 0.2 0.4 0.6 0.8 1

Efficiency & 10kt Background Rate Eff: 0.91, Bkgd: 19.11Hz Eff: 0.88, Bkgd: 5.81Hz Eff: 0.86, Bkgd: 3.87Hz Eff: 0.81, Bkgd: 1.66Hz Eff: 0.70, Bkgd: 0.43Hz Eff: 0.58, Bkgd: 0.10Hz

Efficiency & Background Rates

9

Input to burst trigger:

Different clustering algorithms allow the trade off between lower efficiency and background rate to be explored.

Smaller clusters Larger clusters More background Less background

slide-10
SLIDE 10

Supernova Burst Trigger

10

SN / radiological / noise simulations

Hit-clustering Clustering efficiency & background acceptance different clustering configurations Burst trigger

slide-11
SLIDE 11

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.

slide-12
SLIDE 12

Result: Galactic Neighbourhood Coverage

12

SN Distance, (kpc) 10 20 30 40 50 Efficiency x SN Probability

8 −

10

7 −

10

6 −

10

5 −

10

4 −

10

3 −

10

2 −

10

1 −

10

Galactic Neighbourhood Coverage, Fake Trigger Rate 1/Month

Individual Marley Efficiency & 10kt Background Rate Eff: 0.91, Bkgd: 19.11Hz Eff: 0.88, Bkgd: 5.81Hz Eff: 0.86, Bkgd: 3.87Hz Eff: 0.81, Bkgd: 1.66Hz Eff: 0.70, Bkgd: 0.43Hz Eff: 0.58, Bkgd: 0.10Hz SN Probability

Galactic Neighbourhood Coverage, Fake Trigger Rate 1/Month

LMC Milky Way

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). Assuming a fake burst trigger rate of 1/month, what is our SN sensitivity vs. distance? Close SN, easy to trigger Galaxy edge, harder to trigger

Require larger clusters to increase sensitivity

slide-13
SLIDE 13

Galactic Neighbourhood Coverage 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 Fake Trigger Rate, (Hz)

13 −

10

12 −

10

11 −

10

10 −

10

9 −

10

8 −

10

7 −

10

6 −

10

5 −

10

4 −

10

3 −

10

2 −

10

1 −

10 1 10

Individual Marley Efficiency & 10kt Background Rate Eff: 0.91, Bkgd: 19.11Hz Eff: 0.88, Bkgd: 5.81Hz Eff: 0.86, Bkgd: 3.87Hz Eff: 0.81, Bkgd: 1.66Hz Eff: 0.70, Bkgd: 0.43Hz Eff: 0.58, Bkgd: 0.10Hz

1/Month

Result: Fake Trigger Rate vs. Galactic

13

Trigger on 98% of SN in the neighbourhood, issuing 1 fake trigger per month. In 10 kT Previous slide required 1/month fake trigger rate. This slide shows trade off between efficiency and this rate.

slide-14
SLIDE 14

Time Profile Studies

14

slide-15
SLIDE 15

Recording Length, (s) 5 10 15 20 25 30 35 Coverage 0.2 0.4 0.6 0.8 1

< 1.00 Events Remaining < 0.50 Events Remaining < 0.20 Events Remaining < 0.05 Events Remaining

Supernova Distance, (kpc) 1 10

2

10 Time Since First Neutrino Passed Through Detector, (s) 1 10

< 1.00 Events Remaining < 0.50 Events Remaining < 0.20 Events Remaining < 0.05 Events Remaining

Time Profile Studies

15

High data rate: 1.5 TB / s / 10kt module

Time profile studies potentially very sensitive to SN model.

Establish DAQ requirements: Non-volatile buffer (read-out time). Extract maximal information.

Closer SN, longer time. Further SN, less time.

Record all but 1 event in ~ 28 s, any GNSN.

slide-16
SLIDE 16

Summary

16

  • 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.

BURST TRIGGER: 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.

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.

slide-17
SLIDE 17

Backup Slides

17

slide-18
SLIDE 18

Analysis Details

18

GOAL: Understand supernova triggering efficiencies and corresponding background rates for different 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.

  • 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.

/pnfs/dune/scratch/dunepro/MCC10-Production/SuperNovaSamples/v06_60_00/reco/ snb_timedep_radio_dune10kt_1x2x6

slide-19
SLIDE 19

Why not fast hit yet?

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.

htemp

Entries 307076 Mean 1.409 Std Dev 2.083 1 2 3 4 5 6 7 8 9 10 Generator 20 40 60 80 100 120 140 160 180 200

3

10 × Number of Hits

htemp

Entries 307076 Mean 1.409 Std Dev 2.083

Generator Type, Fast Hit (20ADC)

htemp

Entries 396728 Mean 4.227 Std Dev 0.9519 1 2 3 4 5 6 7 8 9 10 Generator 50 100 150 200 250 300

3

10 × Number of Hits

htemp

Entries 396728 Mean 4.227 Std Dev 0.9519

Generator Type, Gauss Hit

Generator type 0 is ‘noise’.

19

slide-20
SLIDE 20

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.

20

slide-21
SLIDE 21

SN Burst Trigger

21

Trigger rate = total background rate x fraction of Gaussian above cut.

Number of Clusters/Time Window 50 100 150 200 250 300 Trigger Rate, (Hz)

17 −

10

15 −

10

13 −

10

11 −

10

9 −

10

7 −

10

5 −

10

3 −

10

1 −

10 10

Number of Clusters in Time Window Required to Trigger vs. Trigger Rate

Individual Marley Efficiency & 10kt Background Rate Eff: 0.91, Bkgd: 19.11Hz Eff: 0.88, Bkgd: 5.81Hz Eff: 0.86, Bkgd: 3.87Hz Eff: 0.81, Bkgd: 1.66Hz Eff: 0.70, Bkgd: 0.43Hz Eff: 0.58, Bkgd: 0.10Hz

1/Month 1/Week 1/Day

267 clusters by fluctuation

  • nce per month

Metric: On what fraction of galactic SN bursts can we trigger for a given fake rate?

Add trigger threshold to axis label. Ditch titles.

slide-22
SLIDE 22

Number of Events in SN Burst 1 10

2

10

3

10

4

10

5

10 Efficiency 0.2 0.4 0.6 0.8 1

Efficiency vs. Number of Events in SN Burst, Fake Trigger Rate: 1/Month

Individual Marley Efficiency & 10kt Background Rate Eff: 0.91, Bkgd: 19.11Hz Eff: 0.88, Bkgd: 5.81Hz Eff: 0.86, Bkgd: 3.87Hz Eff: 0.81, Bkgd: 1.66Hz Eff: 0.70, Bkgd: 0.43Hz Eff: 0.58, Bkgd: 0.10Hz

Background + Burst

22

SN Burst Efficiency =

∞

X

N

Poisson(µ = Mean background rate + Number of Events in Burst)

Consider SN bursts of size 1->3x10^5 events.

N = 267 clusters/month for black curve.

slide-23
SLIDE 23

Distance to Supernova, (kpc) 1 10

2

10

3

10 Expected Number of Events

2 −

10

1 −

10 1 10

2

10

3

10

4

10

5

10

Expected Number of Events for a Given Supernova Distance

Galactic Center LMC Andromeda Galaxy Far Side Galaxy Near Side

  • K. Scholberg, DUNE collaboration meeting. Sep '16

10kT 40kT

Expected Number of Events for a Given Supernova Distance

Burst -> Distance

23

SN Burst Efficiency vs. Number of Events in Burst (Unoscillated) SN Burst Efficiency vs. Distance to Supernova Rule of thumb: ~30 events at 30 kpc

slide-24
SLIDE 24

Efficiency vs. Distance

24

“Galactic Coverage”

SN Distance, (kpc) 1 10

2

10 Efficiency 0.2 0.4 0.6 0.8 1

Efficiency vs. Distance to SN, Fake Trigger Rate: 1/Month

Individual Marley Efficiency & 10kt Background Rate Eff: 0.91, Bkgd: 19.11Hz Eff: 0.88, Bkgd: 5.81Hz Eff: 0.86, Bkgd: 3.87Hz Eff: 0.81, Bkgd: 1.66Hz Eff: 0.70, Bkgd: 0.43Hz Eff: 0.58, Bkgd: 0.10Hz