Fitting Supernova Spectral Parameters with DUNE Erin Conley On - - PowerPoint PPT Presentation

Fitting Supernova Spectral Parameters with DUNE

Erin Conley On behalf of the DUNE Collaboration April 14, 2019

2

Outline

• Introduction

– The Deep Underground Neutrino Experiment (DUNE) – Supernova neutrinos

• Modeling supernova neutrinos in DUNE

– SNOwGLoBES – MARLEY – Pinched-thermal flux model

• Parameter fitting algorithm

– Studying incorrect detector performance assumptions

• Summary

3

• International experiment for neutrino science (1100+ collaborators!)

– Neutrino oscillation physics, supernova physics, nucleon decay

• Two detectors:

– Near detector on-site at Fermilab – Far detector at Sanford Underground Research Facility (SURF) in South Dakota

• Far detector: world’s largest liquid argon time-projection chamber

(40 kton fiducial mass)

– Ionization electrons drift due to high-voltage electric field – Parallel wire planes create 3D images of particle tracks

www.dunescience.org

4

Supernova Neutrinos in DUNE

• Expect ~3000 neutrino

interaction events in DUNE detector for a 10 kpc SN

– Neutrinos of all flavors carry 99% of core collapse energy – LAr is sensitive to !" (versus water/scintillator which are sensitive to ̅ !")

• DUNE devotes much time

into studying theory, event simulation, reconstruction algorithms, etc. related to supernova physics

Number of SN interactions expected to be seen in DUNE detector

5

Simulating Supernova Neutrino Signals

• SNOwGLoBES:

SuperNova Observatories with GLoBES

– GLoBES: General Long Baseline Experiment Simulator

• Open source event

rate calculation tool

http://phy.duke.edu/~schol/snowglobes/

6

Supernova Flux Model

• Supernova neutrino spectrum AKA

“pinched-thermal form”: ! "# = % "# "#

&

exp − + + 1 "# "#

– "#: Neutrino energy – %: Normalization constant (related to luminosity, .) – "# : Mean neutrino energy – +: Pinching parameter; large + corresponds to more pinched spectrum

• Parameters of interest: ., "# , +

Pinched-thermal for a 10kpc supernova (K. Scholberg) Note: Fluence refers to a time-integrated flux.

7

MARLEY: Model of Argon Reaction Low-Energy Yields

• MARLEY models low-energy

!"CC neutrino interactions

• More sophisticated modeling
• f final state particles
• S. Gardiner (http://www.marleygen.org/)

$% = 16.3 MeV

8

Measuring the Flux Parameters

• Use pinched-thermal flux +

MARLEY modeling to simulate event rates in DUNE detector

• Flux parameters play

significant role in !" event rates

• Develop algorithm to

measure, constrain flux parameters based on SNOwGLoBES event rates

9

( )

1) Test Spectrum !", $% ", &"

…

2) Grid with many different combinations of (!, ⟨$%⟩, &)

Parameter Fitting Algorithm

• Algorithm uses the

following tools:

– “Test spectrum” with given set of pinching parameters !", $% ", &" – Grid of energy spectra containing combinations of (!, $% , &)

• Compute '( value between

test spectrum and all grid spectra; determine best-fit grid element, “sensitivity regions” that constrain parameters

10

Studying Biases due to Incorrect Detector Assumptions

• Test spectrum: data from

supernova as observed by DUNE

• Grids: different DUNE detector

performance assumptions

• Change assumptions for test

spectrum, and for grids, to study effect of mismatched assumptions about detector performance

– Study parameter biases introduced by incorrect assumptions using fractional difference from truth:

• Frac. Diff. = * − *,

*,

11

Studying Effect of Detector Performance Knowledge on Bias:

• Each box corresponds to a unique combination of test

spectrum and grid; diagonal boxes correspond to correct assumptions

• Color scale indicates best-fit parameter fractional

difference from truth

• As assumptions get farther from truth, biases

increase; +30% shift in assumed energy resolution yields ±20% bias on '

Test Spectrum Resolution (Percent) Grid Spectra Resolution (Percent)

12

Summary

• DUNE is preparing to observe supernova neutrinos and

extract as much information as possible

• Parameter fitting algorithm used to understand DUNE’s

ability to constrain supernova flux parameters

– 2D fractional difference plots show bias results from imperfect knowledge of detector parameters; helps quantify how well we need to know these parameters

Backup Slides

14

• Neutrino-argon interaction: argon is

ionized by charged secondary particles

– Scintillation light detected by photon detectors provides timing information

• Charged particles drift toward induction

planes, deposit charge on collection plane wires

• Charge deposited on wire planes

– Reconstructed wire objects (signals for specific particles) – Reconstructed 2D hits (single ionized particles) – Reconstructed 2D clusters (ionization of multiple particles) – Reconstructed 3D objects like tracks, showers, space points

LArTPC Schematic

Liquid Argon Time Projection Chamber

15

Forward Fitting: “Sensitivity”

• Use SNOwGLoBES to generate

binned energy spectra for a given set of pinched-thermal parameters !", $%

", &" → “test spectrum”

• Determine () values for all elements

in grid with many combinations of (!, $% , &)

• Minimize () while profiling over 1 or

2 model parameters

• Form sensitivity regions using cut
• n () values

Example () Map

16

Energy Resolution: Introduction

• Determine how smearing

affects parameter measurements – what if

• ur resolution

assumptions are incorrect?

• Smearing matrices: true

deposited energy from MARLEY + LArSoft; smeared with Gaussian resolution from 0 − 30%

17

Examples of Sensitivity Regions

Notes:

• Here we see

superimposed sensitivity regions + best-fit parameters for one test spectrum input into different grids

• We can see how

the areas change and also how the bias in our best-fit measurements change!

! " (10&' ergs)

• . (MeV)

!