Reconstruction in Water Cherenkov: Using Timing Matt Wetstein - - PowerPoint PPT Presentation

▶
reconstruction in water cherenkov
SMART_READER_LITE
LIVE PREVIEW

Reconstruction in Water Cherenkov: Using Timing Matt Wetstein - - PowerPoint PPT Presentation

Feb 01, 2023 220 likes •719 views

University of Chicago Reconstruction in Water Cherenkov: Using Timing Matt Wetstein Enrico Fermi Institute, University of Chicago Argonne National Laboratory on behalf of the LAPPD, LBNE collaborations, and the Fast-Timing Reconstruction

slide-1
SLIDE 1

University of Chicago

Reconstruction in Water Cherenkov:

Using Timing

Matt Wetstein

Enrico Fermi Institute, University of Chicago Argonne National Laboratory

  • n behalf of the LAPPD, LBNE collaborations, and

the Fast-Timing Reconstruction Group:

  • Z. Djurcic (ANL), G. Davies (Iowa State), H. Frisch (U Chicago)
  • M. Sanchez (Iowa/ANL), M. Wetstein (U Chicago/ANL), T. Xin (Iowa State)

Advances in Neutrino Technology - ANT 11 October 12, 2011

Wednesday, October 12, 2011

slide-2
SLIDE 2

University of Chicago

Reconstruction in Water Cherenkov:

Using Correlated Timing and Spatial Information

Matt Wetstein

Enrico Fermi Institute, University of Chicago Argonne National Laboratory

  • n behalf of the LAPPD, LBNE collaborations, and

the Fast-Timing Reconstruction Group:

  • Z. Djurcic (ANL), G. Davies (Iowa State), H. Frisch (U Chicago)
  • M. Sanchez (Iowa/ANL), M. Wetstein (U Chicago/ANL), T. Xin (Iowa State)

Advances in Neutrino Technology - ANT 11 October 12, 2011

Wednesday, October 12, 2011

slide-3
SLIDE 3

ANT 2011 3

Pi0s and efgects on long-baseline physics

π0 π0 π0 π0

boost boost

γ γ γ

after boost, the second low E gamma is too small to reconstruct two forward gammas with angular separations typically smaller than 15º get mis-identified as a single electron shower

Largest reducible background. In WC, in order to achieve a pure electron sample (~1% π0), one needs harsh quality cuts that bring signal effjciency down to 16% (28%) at 1 GeV (0.8 MeV). This loss of events explains the factor of 3-5 larger fiducial mass necessary to match the performance of an LAr There is still a room for significant improvement in the physics capabilities for a given mass of water.

Wednesday, October 12, 2011

slide-4
SLIDE 4

ANT 2011 4

Thinking about Full Track Reconstruction: WC as a Time Projection Chamber?

  • 1. Signal per unit length
  • 2. Drift time
  • 3. Topology

~20 photons/mm ~225,000mm/microsecond drift distances depend

  • n track parameters

Wednesday, October 12, 2011

slide-5
SLIDE 5

ANT 2011 5

  • 1. Signal per unit length
  • 2. Drift time
  • 3. Topology

Acceptance and coverage are important, especially at Low E. Is there any way we can boost this number? Scintillation?

Thinking about Full Track Reconstruction: WC as a Time Projection Chamber?

This necessitates fast

  • photodetection. It also requires

spatial resolution commensurate with the time resolution. This presents some reconstruction challenges...

~20 photons/mm ~225,000mm/microsecond drift distances depend

  • n track parameters

Wednesday, October 12, 2011

slide-6
SLIDE 6

ANT 2011 6

LAPPD (Large-Area Picosecond Photodetector) Project:

Make large-area MCPs with low-cost, bulk materials and batch industrial techniques

  • We’re attacking all

aspects of this problem from the photocathode to the MCPs to vacuum sealing technology

  • Goal is not just proof of

principle...It’s the development of a commercializable product.

!"#$%# &"#$%#

IJ&.<& KJ&.<&

New Developments in Water-Based Detectors: Large Area, High Resolution MCP-PMTs

Wednesday, October 12, 2011

slide-7
SLIDE 7

ANT 2011 7

4 5 6 7 8 9 10 11 12 13 x 10

9

500 1000 1500 2000 2500

Arrival Time (seconds)

Transit Time Spread, MCP 72/78 at 2.6kV with 1kV across anode gap

Transit Time Spread for MCP 72/78 at 2.6 k

arrival time (nanoseconds)

1 10 100 1000 10000 100000 600 750 900 1050 1200 ALD-MCP 122 ALD-MCP 125-133 Commercial MCP field-strength (Volts/mm) average gain

6 6.5 7 7.5 8 8.5 x 10

9

0.05 0.05 0.1 0.15 time (seconds) voltage (Volts) 300 V across the PC gap 100 V across the PC gap 60 V across the PC gap 20 V across the PC gap

Wetstein (U Chicago/ANL-HEP), B. Adams, M. Chollet(ANL-APS)

New Developments in Water-Based Detectors New Developments in Water-Based Detectors: Large Area, High Resolution MCP-PMTs

725000 893750 1062500 1231250 1400000 100 200 300 400 Data, 1.5 kV Data 1.36 kV Simulation, 1.5 kV Simulation, 1.44 kV Simulation, 1.36 kV

Wednesday, October 12, 2011

slide-8
SLIDE 8

ANT 2011 8

From SciBoone to....DanielBoone?

courtesy of Zelimir Djurcic

DANIELBOONE

Next Generation Water Cherenkov Experiment

ν

  • expected rate of ~90k CC events and ~35k NC

events per 1E20 POT, 10.6 ton fiducial volume

  • naively scale the event rate to oxygen using a

factor 8/6, then you'd expect ~12.4k CC interactions/ton/1E20 and ~4.7k NC interactions/ ton/1E20

  • Booster Neutrino Beam rate is ~0.5E19 POT per

week on average

  • ~32.2k CC interactions/ton/year, ~11k NC

interactions/ton/year in a water detector BEFORE accounting for the efficiencies of such a detector.

Concept and event rate calculation, courtesy of David Schmitz, FNAL

First step: to demonstrate a working, small-scale neutrino detector

Wednesday, October 12, 2011

slide-9
SLIDE 9

ANT 2011 9

Beyond DanielBoone...How do you scale this technology up?

~1m3!

~20 ton! ~100 ktons!

Cost isn’t just number of PMTs excavation costs, magnetic shielding (not necessary for MCPs), more or better use fiducial volume, competition in the PMT market In particular, lower costs and more effjcient use of target mass means leaves room for departure from conventional WC design in ways that could enhance analysis

  • possibility of segmentation? Less losses, scattering, dispersion...
  • applied magnetic fields?

Disclaimer: This is not targeted at LBNE stage I: possible upgrade path or second detector....Or something beyond...

Building a large Water detector, based on LAPPDs is not simply a matter of building a Super K type detector with MCPs

  • Difgerent optimization of volume to length scale
  • Difgerent balance of cost per surface area
  • Difgerent physics capabilities for the same fiducial mass

Lower costs are important, but physics reach should be the focus

Wednesday, October 12, 2011

slide-10
SLIDE 10

ANT 2011 10

  • These new capabilities and new possible
  • ptimizations necessitate a strong simulation and

reconstruction program.

  • Collaboration among the hi-res WCh working

group has produced a new platform for testing algorithms on WCh detectors with interactively modifiable photodetector properties.

  • These efgorts have already identified promising

features in observables, such as timing residuals, that could potentially be used to improve track reconstruction and better identify pi0 backgrounds.

  • GEANT-based studies are ongoing...there is much

work ahead. Plans for new post-docs and students to be starting soon.

  • These efgorts are well integrated into the WC

algorithms development efgorts for LBNE. Many

  • f the tools will be useful for the nominal LBNE

design.

trackresidualdist

Entries 29230 Mean 3.184 RMS 3.601 / ndf

2

  • 597.4 / 417

Constant 9.7 ! 1096 MPV 0.009 ! 1.278 Sigma 0.0048 ! 0.6612

  • 10
  • 5

5 10 15 20 25 20 40 60 80 100 120 140 160 180 200 trackresidualdist

Entries 29230 Mean 3.184 RMS 3.601 / ndf

2

  • 597.4 / 417

Constant 9.7 ! 1096 MPV 0.009 ! 1.278 Sigma 0.0048 ! 0.6612

Rising edge driven by chromatic dispersion Tail driven by scattered light

HiRes Water Group

  • Z. Djurcic (ANL), G. Davies (Iowa State), H. Frisch (U Chicago)
  • M. Sanchez (Iowa/ANL), M. Wetstein (U Chicago/ANL),
  • T. Xin (Iowa State)

timing residual (ns)

  • 5

5 10 0.0005 0.001 0.0015 0.002 0.0025

Wednesday, October 12, 2011

slide-11
SLIDE 11

Can we use slices in constant time to fully reconstruct tracks?

ANT 2011 11

Isochrons electron pi0

Wednesday, October 12, 2011

slide-12
SLIDE 12

ANT 2011 12

s1 s2 α Δt ≈ s1/c + s2 n/c Connect each hit to the vertex, through a two segment path, one segment representing the path of the charged particle, the other path representing the emitted light. There are two unknowns: s1 and α but there are two constraints: s1 + s1 = d and Δtmeasured = s1/c + s2 n/c d Track Reconstruction Using an “Isochron Transform” θc The isochron method is a Hough Transform in 4- space, that builds tracks from a pattern of hits in time and space.

Wednesday, October 12, 2011

slide-13
SLIDE 13

ANT 2011 13

200 400 600 800 1000 100 200 300 400 500 600 700 800 900 1000

Trajectories of Constant Transit Time

And of course, there are a degenerate number of track directions, from which a photon emitted at the Cherenkov angle can hit Track Reconstruction Using an “Isochron Transform”

Wednesday, October 12, 2011

slide-14
SLIDE 14

ANT 2011 14

However multiple hits from the same point of emission, will maximally intersect along the point of emission...This is similar to the Hough transform. Track Reconstruction Using an “Isochron Transform”

Wednesday, October 12, 2011

slide-15
SLIDE 15

200 400 600 800 1000

  • 400
  • 200

200 400 600 2000 4000 6000 8000 1000 1200 1400 1600 1800

sohp

ANT 2011 15 200 400 600 800 1000

  • 400
  • 200

200 400 2000 4000 6000 8000 1000 1200 1400 1600 1800 2000 2200 2400

sohp

Results of a toy Monte Carlo with perfect resolution

Color scale shows the likelihood that light on the Cherenkov ring came from a particular point in space. Concentration of red and yellow pixels cluster around likely tracks

Single track Two tracks displaced from a common vertex Track Reconstruction Using an “Isochron Transform”

Wednesday, October 12, 2011

slide-16
SLIDE 16

ANT 2011

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

200 400 600 800 1000 1200 1400 1600 1800 2000 2200

Reconstructed 750 MeV Muon (geant) 16

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500

Reconstructed 750 MeV Electron (geant)

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500 4000

Reconstructed 750 MeV Pi0 (geant)

  • Events were generated at the center of an

6m cube.

  • These are just the crude (hand-tweeked)

isochrons, constructed with respect to the primary vertex.

  • Further steps could improve these

transforms.

  • Nonetheless, contrast between the

particles, is still pretty stark.

Reconstructing Geant Events

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500

Reconstructed 750 MeV Pi0 (geant)

Wednesday, October 12, 2011

slide-17
SLIDE 17

ANT 2011

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

200 400 600 800 1000 1200 1400 1600 1800 2000 2200

Reconstructed 750 MeV Muon (geant) 17

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500

Reconstructed 750 MeV Electron (geant)

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500 4000

Reconstructed 750 MeV Pi0 (geant)

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500

Reconstructed 750 MeV Pi0 (geant)

Comparing Isochron Reconstruction.... If I hand draw track hypotheses through these transforms...

Wednesday, October 12, 2011

slide-18
SLIDE 18

ANT 2011 18

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500 4000

Reconstructed 750 MeV Pi0 (geant)

x position (mm) 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500 500 1000 1500 2000 2500 Emitted Photons Along Muon Track (geant-truth) x position (mm) 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500 1000 2000 3000 4000 5000 6000 7000 8000 9000 Emitted Photons Along Electron Track (geant-truth) x position (mm) 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500 1000 2000 3000 4000 5000 Emitted Photons Along Pi0 Track (geant-truth)

With True Tracks They match very nicely with the truth-level tracks/shower constituents

Wednesday, October 12, 2011

slide-19
SLIDE 19

ANT 2011 19

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500

Reconstructed 750 MeV Electron (geant)

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500 4000

Reconstructed 750 MeV Pi0 (geant)

x position (mm) 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500 500 1000 1500 2000 2500 Emitted Photons Along Muon Track (geant-truth) x position (mm) 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500 1000 2000 3000 4000 5000 Emitted Photons Along Pi0 Track (geant-truth)

With True Tracks

x position (mm) 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500 1000 2000 3000 4000 5000 6000 7000 8000 9000 Emitted Photons Along Electron Track (geant-truth)

They match very nicely with the truth-level tracks/shower constituents

Wednesday, October 12, 2011

slide-20
SLIDE 20

ANT 2011 20

Optimizing the Transform

Once candidate showers have been identified, each stage of the shower needs to be independently transformed. Isochron algorithm works best over one single stage of a shower. Each new branch point can be transformed iteratively, using the branch point as the new starting vertex. Crude, first application of the isochron transform is useful to identify the original vertex, location of first light, number of shower candidates in the next stage...But, these are important particle ID handles. Can be used to make initial cuts. In this first stage, before later applying chromatic corrections, we can very the index of refraction to look for shower candidates....

1. 2.

Wednesday, October 12, 2011

slide-21
SLIDE 21

500 1000 1500 2000

  • 1000

1000 500 1000 1500 2000 2500 3000

proj_0

500 1000 1500 2000

  • 1000

1000 1000 2000 3000 4000 5000 6000

proj_19

ANT 2011 21

Isochron Transform Evaluated for Difgerent Wavelengths

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500

Reconstructed 750 MeV Pi0 (geant)

As a first pass, on can adjust the default refractive index used by the transform in order to optimize the track fit. This is equivalent to varying the ring size in a conventional super K Hough transform... Instead of searching for a second peak, Pi0 identification consists of searching for a second track-like

  • bject as we vary N.

Wednesday, October 12, 2011

slide-22
SLIDE 22

ANT 2011 22

Understanding Chromatic Dispersion

  • A concern in using fast timing are the

efgects of frequency dependent dispersion, scattering and absorption.

  • Using a fast toy MC originally developed

by J. Felde we study the time of arrival for photons in an spherical detector.

  • For a 50m detector with 100% coverage,

the rise time (t90-t10) is of the order of 2 ns which cannot be sampled with standard PMT technology.

  • For a given detector size, the rise time

stays constant and the uncertainty in the position of the leading edge becomes smaller if larger photodetector coverage is considered.

  • A combined improvement in

photodetector coverage (for reduced uncertainty in risetime) and faster timing (to better sample the risetime) allows for better use of timing information in Water Cherenkov detectors.

/ ndf

2

  • 63.95 / 29

Prob 0.0001943 Constant 4.4 ! 240.5 Mean 0.1 ! 366.4 Sigma 0.038 ! 1.756

Time (ns)

360 365 370 375

Photons

50 100 150 200 250

/ ndf

2

  • 63.95 / 29

Prob 0.0001943 Constant 4.4 ! 240.5 Mean 0.1 ! 366.4 Sigma 0.038 ! 1.756

Photon time of flight

150 300 450 600 20 50 80

Uncertainty on Arrival Time

Uncertainty (psec) Distance from Detector (meters) 100% Coverage 30% Coverage 10% Coverage

  • J. Felde, B. Svoboda UC Davis

Wednesday, October 12, 2011

slide-23
SLIDE 23

200 400 600 800 1000

  • 400
  • 200

200 400 600 2000 4000 6000 8000 1000 1200 1400 1600 1800

sohp

ANT 2011 23

Once fit with a track, light color is fully determined.

An Example Analysis

trajectory for correct color will intersect with the track trajectory for incorrect color will not align with the rest of the track

Given known absorption properties, and Cherenkov spectra, it should also be possible to apply a Bayesian unfolding method to reverse the efgects of chromatic dispersion, and sharpen the track features. However, this is computation-intensive and should only be applied as a second step. We’re working on the best way to do this....

Wednesday, October 12, 2011

slide-24
SLIDE 24

ANT 2011 24

Understanding Chromatic Dispersion Δs D

(distance scale)

Light of difgerent colors arrives at difgerent times, but also at difgerent Cherenkov angles. At 10 meters distance, 250nm light arrives ~6.5 nsec later than 550 nm. Given the difgerence in θc for the two colors (0.885 versus 0.747 radians), the spatial separation between the red and blue light at 10 meters is ~1.4 meters with resolution on Δt approaching 100 picoseconds, one can distinguish between colors much closer on the spectrum, but only if

  • ne can also resolve the

corresponding Δs which is 2.4 cm in LBNE, granularity of PMTs is 10” (25.4 cm) with more the 1 meter separation between phototubes.

Wednesday, October 12, 2011

slide-25
SLIDE 25

x vertex (mm)

  • 20
  • 15
  • 10
  • 5

5 10 15 20 z vertex (mm)

  • 20
  • 15
  • 10
  • 5

5 10 15 20 480 500 520 540 560 580 600 620

Avg No of Intersections per Bin in Bins w >550 Intersections 200 400 600 800 1000 1200 1400 1 10

2

10

3

10

4

10

5

10

ANT 2011 25

An Example Analysis

Sharpness of intersection points can be used as a figure-of-merit for fitting vertices This allows you to work back to the original vertex that causally unites all of the detector hits...even if that common vertex proceeds the onset of Cherenkov light... Later branch-points in the EM shower will appear as local minima....

Wednesday, October 12, 2011

slide-26
SLIDE 26

x-position (mm)

  • 2000
  • 1500
  • 1000
  • 500

500 1000 1500 2000 z-position (mm)

  • 2000
  • 1500
  • 1000
  • 500

500 1000 1500 2000 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

Vertex Likelihood

x position (mm)

  • 2000
  • 1000

1000 2000 5000 10000 15000 Light Emmission Profile Projected on x-axis ANT 2011 26

Vertex Separation As A Handle on PID

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500

Reconstructed 750 MeV Pi0 (geant)

We have one strong vertex that causally connects the hit pattern, located far before the

  • nset of Cherenkov light. This

is followed by two vertex candidates, corresponding roughly with first light. First light and vertex candidates correspond nicely with the two shower candidates reconstructed by the isochron method

Pi0

Location of the candidate vertices do not correspond perfectly to true vertices, probably due to an inappropriate choice of refractive index. Wednesday, October 12, 2011

slide-27
SLIDE 27

x-position (mm)

  • 2000
  • 1500
  • 1000
  • 500

500 1000 1500 2000 z-position (mm)

  • 2000
  • 1500
  • 1000
  • 500

500 1000 1500 2000 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

Vertex Likelihood

x position (mm)

  • 2000
  • 1000

1000 2000 5000 10000 15000 Light Emmission Profile Projected on x-axis x-position (mm)

  • 2000
  • 1000

1000 2000 z-position (mm)

  • 2000
  • 1000

1000 2000

0.5 1 1.5 2 2.5 3 3.5

Vertex Likelihood ANT 2011 27

Vertex Separation As A Handle on PID

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500

Reconstructed 750 MeV Pi0 (geant)

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500

Reconstructed 750 MeV Electron (geant)

x-position (mm)

  • 2000
  • 1000

1000 2000

z-position (mm)

  • 2000
  • 1000

1000 2000

0.5 1 1.5 2 2.5 3 3.5 4

Vertex Likelihood

x position (mm)

  • 2000
  • 1000

1000 2000 5000 10000 15000 Light Emmission Profile Projected on x-axis

We have one strong vertex that causally connects the hit pattern, located far before the

  • nset of Cherenkov light. This

is followed by two vertex candidates, corresponding roughly with first light. First light and vertex candidates correspond nicely with the two shower candidates reconstructed by the isochron method

Electron

Location of the candidate vertices do not correspond perfectly to true vertex, probably due to an inappropriate choice of refractive index (I used the wrong value). Wednesday, October 12, 2011

slide-28
SLIDE 28

ANT 2011 28

Conclusions

  • Discrimination between pi0s and electrons at higher energies (>1 GeV) in WC is a diffjcult

task, but with high pay ofg for long baseline neutrino physics: We have a factor of ~5 loss we can recover...

  • Advances in photodetection technology provide us with a new opportunity re-imagine

water-based neutrino detectors.

  • Simultaneous use of fine binned timing and high spatial resolution can provide enough

information to think about close-to-full event reconstruction in water, but necessitate new analysis techniques.

  • Chromatic dispersion is a confounding factor, making analysis more diffjcult..but it can

be corrected for.

  • Attenuation and scattering losses (as well as low light yield) are the biggest challenges for

a large detector. These could be optimized (novel design) or mitigated (water-based scintillator- M. Yeh, BNL).

  • This talk focused on accelerator neutrinos, but there are countless other applications. We

need your help imagining applications for these tools.

x position (mm) 500 1000 1500 2000 z position (mm)

  • 1500
  • 1000
  • 500

500 1000 1500

500 1000 1500 2000 2500 3000 3500

Reconstructed 750 MeV Pi0 (geant)

Wednesday, October 12, 2011

slide-29
SLIDE 29

ANT 2011 29

Thank you

to Henry Frisch and Mayly Sanchez for their help with the talk Thanks to Conference Organizers for making ANT happen to all of my LBNE and LAPPD colleagues for all of the work presented in this talk Wednesday, October 12, 2011

slide-30
SLIDE 30

ANT 2011 30

Backup Slides

Wednesday, October 12, 2011

slide-31
SLIDE 31

ANT 2011 31

Next Steps

  • Developing an efgective reconstruction recipe.
  • Implement explicit track and vertex fitting based on isochron transforms.
  • Development of tools for correcting chromatic dispersion
  • Develop metrics for particle ID. Test those metrics.
  • Studying algorithm performance over variations in drift distance
  • Apply isochron algorithm on LBNE (WCSim) simulations.
  • Make the code fast and effjcient.

Wednesday, October 12, 2011

slide-32
SLIDE 32

ANT 2011 32

Neutrino Interactions with Matter Charged Current (signal) Neutral Current (bkgd)

W± l± Z0 Recoil(π0) νl

Typical neutrino oscillation experiments detect neutrinos through their interactions with matter. Specifically neutrino flavor can be determined by charged-current interactions, which produce charged leptons of like flavor. The ability to discriminate between types of neutrinos is then limited by the ability of a detector paradigm to discriminate between high energy particles in the fiducial volume.

Wednesday, October 12, 2011

slide-33
SLIDE 33

ANT 2011 33

Intro to WC Detectors

Credit: Mark Messier

  • An shockwave of optical

light is produced when a charged particle travels through a dielectric medium faster than the speed of light in that medium: c/n

  • This light propagates at

an angle θC = acos(1/nβ) with respect to the direction of the charged particle…

  • Using photodetectors, we

can measure the positions and arrival times of the emitted light and reconstruct the track

Wednesday, October 12, 2011

slide-34
SLIDE 34

ANT 2011 34

electron pi0

Wednesday, October 12, 2011

slide-35
SLIDE 35

ANT 2011 35

electron pi0

Wednesday, October 12, 2011

slide-36
SLIDE 36

wavelength (nm) 200 250 300 350 400 450 500 550 speed of light (m/sec) 170 180 190 200 210 220

6

10 ×

wavelength (nm) 200 250 300 350 400 450 500 550 600 0.002 0.004 0.006 0.008 0.01 0.012

Spectrum of Direct Light Traversing 25 m (geant)

ANT 2011 36

Wednesday, October 12, 2011

slide-37
SLIDE 37

ANT 2011 37

Timing Information: π0

~1 radiation length ~37 cm vertices are separated: at 7 degrees: ~4.5 cm at 15 degrees: ~9.7 cm

On average, this amounts to separating the two vertices from which the Cherenkov cones radiate...

  • Finding a single event vertex is limited by
  • ur ignorance of T0.
  • Vertex separation is not...

Wednesday, October 12, 2011

slide-38
SLIDE 38

ANT 2011 38

Timing Information: π0

in term of time: at 7 degrees: ~200 psec at 15 degrees: ~425 psec

  • Finding a single event vertex is limited by
  • ur ignorance of T0.
  • Vertex separation is not...

speed of light in water: ~44 psec/ cm

~1 radiation length ~1.64 nsec

On average, this amounts to separating the two vertices from which the Cherenkov cones radiate...

Wednesday, October 12, 2011

slide-39
SLIDE 39

ANT 2011 39

New Developments in Water-Based Detectors: Possibility of Water-Based Scintillator Linear Alkylbenzene (LAB) - Industrial detergent Key innovations:

  • ability to create stable solutions
  • purification to achieve longer attenuation lengths

Ideal for large scale experiments

  • Non-toxic
  • Non-flammable
  • Stable
  • Cheap

The scintillation light might be diffjcult to resolve with timing, but...

  • It may be possible to have both Cherenkov and

scintillation light, separated in time

  • The spatial/statistical gains would be

considerable. Minfang Yeh et al, Brookhaven National Lab

This slide is courtesy of M. Yeh. Special thanks also to Howard Nicholson.

Wednesday, October 12, 2011

slide-40
SLIDE 40

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11 40

Several Approaches To WC reconstruction

  • Use pure timing of hits
  • Use pattern-of-light fitting, based on Geant simulations of light yields and

transmission

  • Use geometric properties of Cherenkov light, combined with timing

Wednesday, October 12, 2011

slide-41
SLIDE 41

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11 41

Several Approaches To WC reconstruction

  • Use pure timing of hits
  • Use geometric properties of Cherenkov light, combined with timing

Need to solve systems of equations for 7 degrees of freedom for septuplets of hits. Combinatorics become hairy...but worth pursuing

  • Use pattern-of-light fitting, based on Geant simulations of light yields and

transmission Best way to do things with brute force computing, not the focus of this talk but will be a major part of this effort. Best way to conceptually understand the problem, focus of this talk...

Wednesday, October 12, 2011

slide-42
SLIDE 42

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11 42

Track Reconstruction

∆thyp = ns/c0

s = | D - v | v=(xv,yv,zv,T0.)

Step 1:

Conceptualize Cherenkov light as coming from a point source…

  • For different hypothesized point-

source locations compute the distance (s) between the point and the detector hits.

  • Calculate the hypothesized time

(∆thyp) it would take the light to reach each detector from the point source.

  • Adjust the location of the point source

to minimize the width of the (∆thyp-tD)2/σt distribution for all hits. Fit pameters: xv,yv,zv,T0

“Simple Vertex”

Wednesday, October 12, 2011

slide-43
SLIDE 43

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11 43

Track Reconstruction

∆thyp = s1/c0 + ns2/c0 7 Fit pameters: xv,yv,zv,T0,θtrack,φtrack,β 0-D Point Source 1-D Track

s1 v=(xv,yv,zv,T0.) s2

“Extended Track”

Wednesday, October 12, 2011

slide-44
SLIDE 44

track time residual (ns)

20 40 100 200 300 400

track theta (radians)

  • 1.5
  • 1
  • 0.5

100 200 300 400 500

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11 44

Basic Observables Used in Fits

position of emitted light along track (cm)

  • 20000
  • 10000

10000 20000 20 40 60

arrival time (ns)

100 200 300 100 200 300

Wednesday, October 12, 2011

slide-45
SLIDE 45

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11 45

Track Reconstruction

richer structure 1-D Track multi-scattered muon

Wednesday, October 12, 2011

slide-46
SLIDE 46

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11 46

Track Reconstruction

richer structure 1-D Track a showering particle

Wednesday, October 12, 2011

slide-47
SLIDE 47

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11 47

richer structure 1-D Track two showering particles

Track Reconstruction

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11

Wednesday, October 12, 2011

slide-48
SLIDE 48

LAPPD Collaboration: Large Area Picosecond Photodetectors UChicago - HEP Lunch Seminar 03/14/11 48

Ring Counting Particle ID

\

Hough transformation converts ring counting into peak counting

Wednesday, October 12, 2011