Measuring High-Energy -Ray Spectra with HAWC Sam Marinelli for the - - PowerPoint PPT Presentation

Measuring High-Energy γ-Ray Spectra with HAWC

Sam Marinelli for the HAWC Collaboration

Michigan State University

August 9, 2017

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 1 / 13

The High-Altitude Water-Cherenkov observatory

Detects TeV γ rays at 4100 m on the Sierra Negra mountain in Puebla, Mexico. 1200 PMTs in 300 water-filled tanks detect Cherenkov light from air showers. Timing used to determine shower direction.

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 2 / 13

HAWC tanks

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 3 / 13

HAWC energy estimation via artificial neural network (NN)

Using Toolkit for Multivariate Analysis1. NN maps several event-wise variables to estimate of primary energy. 479 free parameters chosen by training on Monte Carlo (MC).

1

x

2

x

3

x

11

w 1

12

w 1 y2

1

y2

2

y2

3

y2

4

y2

5

y1

3

y1

1

y1

2

y1

4

y3

1 11

w 2

51

w 2

01

w 2

05

w 1

45

w 1

4

x

Input Layer Output Layer Hidden Layer

Bias Bias y

ANN

1 1

1http://tmva.sourceforge.net/.

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 4 / 13

NN input variables

Input variables chosen to characterize shower size and geometry. Shower characteristic Input variables Energy deposited in the detec- tor Fraction of PMTs hit Fraction of tanks hit Normalization of lateral-distribution fit Fraction

• f

ground energy landing on the detector Distance of shower core from detector center Fraction

• f

primary energy reaching the ground Zenith angle Lateral energy distribution

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 5 / 13

Performance on simulation

NN energy better correlated with MC truth than currently used variable (fraction of PMTs hit). Ability to determine energies beyond 100 TeV.

Fraction of Events

9 −

10

8 −

10

7 −

10

6 −

10

5 −

10

4 −

10

3 −

10

2 −

10 [True energy (eV)]

10

log 10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15 [Reconstructed energy (eV)]

10

log 10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15

NN energy

Fraction of Events

9 −

10

8 −

10

7 −

10

6 −

10

5 −

10

4 −

10

3 −

10

2 −

10 [True energy (eV)]

10

log 10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15

hit

f

10

log 1.2 − 1 − 0.8 − 0.6 − 0.4 − 0.2 −

Fraction of PMTs hit

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 6 / 13

RMS error

RMS error of ∼32% at highest energies. Use of lateral distribution compensates for fluctuations in height of first interaction. Other techniques Ground Parameter energy- reconstruction method described by Kelly Malone

• n August 8 at

15:00.

[True energy (eV)]

10

log 11 11.5 12 12.5 13 13.5 14 14.5 Log-energy RMS error 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Neural Net Likelihood Ground Parameter

~20% ~40% ~60% ~80% ~100%

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 7 / 13

HAWC Crab spectrum using NN

Events binned two-dimensionally in fraction of PMTs hit and NN energy. Poisson-likelihood forward-folded fit is applied to these bin contents. Crab modeled as point source with log-parabola γ-ray spectrum: dN dE = k (E/E0)−α−β ln(E/E0) . (1) Fit serves as proof of principle for energy reconstruction but may also constrain high-energy Crab physics.

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 8 / 13

Crab fit result

Statistical errors using new energy variables are smaller than in published HAWC result2. Systematics analysis in

• progress. Assuming 50%

flux systematic from published analysis, fits with new energy variables are compatible with H.E.S.S. measurement.

100 101 102 E (TeV) 10-13 10-12 10-11 10-10 E 2dN/dE (TeV/cm2/s)

Preliminary Neural Net Ground Parameter H.E.S.S. ICRC 2017

Dark band – statistical error Light band – systematic error

2https://arxiv.org/abs/1701.01778.

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 9 / 13

HEGRA Crab Nebula spectrum (Aharonian et al. 2014)

• Stat. errors at highest energies

comparable to HEGRA’s. Might be improved with tuned cuts.

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 10 / 13

Implications of measurement for PWN modeling

Interpretation of HAWC result requires understanding at what energies spectrum is being measured. High-energy γ spectrum sensitive to highest-energy electron acceleration. Models De Jager et al. model PWN high-energy inverse-Compton emission. Atoyan and Aharonian (1995) also suggest bremsstrahlung could play a role if PWN inhomogeneous.

De Jager et al. (1996).

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 11 / 13

Sensitivity to Lorentz-invariance violation

Lorentz- invariance- violating models predict γ decay to e+e− above some energy. Detection of high-energy γ rays constrains this energy scale. HAWC Crab spectrum will imply some limit.

Mart´ ınez-Huerta and P´ erez-Lorenzana (2017).

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 12 / 13

Future work

Crab analysis not yet optimized. Must tune cuts etc. to new spectral-fitting technique. Galactic plane in 56–100 TeV map, made with 1◦ extended-source model and assuming 2.63 spectral index, shows several known sources. With new energy variables, HAWC can attempt measurements of these sources’ spectra at unprecedented energies.

7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 l [ ◦ ] 2 1 1 2 b [ ◦ ]

PRELIMINARY

1 2 3 4 5 6 Significance

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 13 / 13

Bonus round

Backup slides.

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 14 / 13

Sensitivity to time variability

Mart´ ın et al. numerically models time dependence of spinning down pulsar/PWN. Cooling time for PeV electrons is ∼1 month. HAWC could look for spectral variations on this time scale.

Mart´ ın et al. (2012).

• S. S. Marinelli (MSU)

High-Energy Spectra with HAWC August 9, 2017 15 / 13