Conversion of MAGIC data to DL3 format L. Jouvin, J. Delgado, C. - - PowerPoint PPT Presentation

1

Conversion of MAGIC data to DL3 format

• L. Jouvin, J. Delgado, C. Nigro, J. Rico

2

MAGIC

• At the Roque de los Muchachos Observatory in La Palma
• Operating since 2004
• Since 2009: 2 Imaging Atmospheric Cherenkov telescopes
• 17 m diameters
• Summers 2011-2012: telescopes underwent a series of upgrades
• Sensitive between 50 GeV and 30 TeV

3

DL3 format

DL3 is event lists + IRFs for each observation after applying analysis cuts.

• Events list: γ-like events
• IRFs:
• Effective area
• Energy resolution
• PSF
• Background
• Two types of IRF:

– Pointlike – Full

θ2

θ0

4

First joint analysis of the Crab Nebula

• DL3 MAGIC converter
• Perform a multi-instrument spectral analysis of the Crab Nebula with Gammapy
• C. Nigro et al.

5

3D likelihood analysis

• Source morphology: more and more precise and complex
• This complexity will increase with CTA observation
• cube analysis: fit simultaneously the morphology and the

spectrum on a cube dataset → required to separate the different components of a same region of the sky

For morphological 2D analysis and 3D analysis, we need: → IRF full enclosure (without theta2 cut) → Offset dependant in the FOV → PSF → Background model

6

MAGIC analysis software (MARS)

7

DL3 Event List

• Melibea level in MARS contain:
• Incident Direction
• Energy of the primary gamma-ray
• Hadronness (gamma/hadrons separation)
• Good time interval
• Created for each run (~20min) of observations, or for MC events.
• Fix a hadronness → event list for this cut
• DL3 level for event lists is the output of the Melibea program

8

DL3 IRF

• flute and the DL3 exporter
• flute:

–

Start from data and MC from the melibea level.

–

Weight IRF by zenith and azimuth distribution of the events → Produce IRF for each observation condition.

–

Used to compute:

➢ Effective area in true energy ➢ The effective on time in each bin of azimuth and zenith ➢ List of MC events containing: true and reconstructed energy

• Energy dispersion in DL3 exporter: list of MC events weighting by

the effective on time distribution for each run

• Use single offset MC simulation at 0.4°
• Pointlike

9

Main current work in the DL3 group

• General pipeline
• Offset dependent IRF in the FOV
• PSF for each observation
• Full IRF

10

General pipeline for all MC period

• Summer 2011-2012: telescopes underwent a series of upgrades
• 8 MC periods → we want to produce general RFs that covers each

specific MC period

• deliver complete DL3 release for each MC period
• started with one MC period : 2013-07-27 - 2014-06-18
• Most of the observations: pointlike source at 0.4°

11

Start with one MC period

• under development with Jordi Delgado
• 4 known pointlike sources

General pipeline phi1TeV = (4.22 +/- 0.30) 10-11 cm-2 TeV-1 s-1 Alpha= 2.52 +/- 0.08 Beta= 0.16 +/- 0.04 Specific RF phi1TeV = (4.14 +/- 0.30) 10-11 cm-2 TeV-1 s-1 α= 2.55 +/- 0.09 β= 0.17 +/- 0.05

Joint Crab paper: 2 runs of 04/10/2013

12

Offset dependent IRF

13

From diffuse MC

• Use diffuse MC
• creating rings at different offset

Select a specific ring

Diffuse Ring MC

0.5° 0.3°

→ IRF at 0.4° Different Ring → IRF offset dependant in the FOV

14

Crab offset dependant observation

MC period: 2013-07-27 – 2014-06-18

Diffuse ring:

• 0 – 0.3 °
• 0.3 – 0.5 °
• 0.3 – 0.5 °
• 0.5 – 0.8 °
• 0.8 – 1.1 °
• 1.3 – 1.4 °

Offset:

• 0.2 °
• 0.35 °
• 0.4 °
• 0.7 °
• 1.0 °
• 1.4 °

15

Crab Nebula spectrum at different

• ffset

→ Large offset: camera inhomogeneties, not possible to use reflected region → MAGIC software: correction by the acceptance possible for the background extraction. → Problem with Gammapy that uses only reflected region

16

Flute and DL3 exporter produce IRFs offset dependant

Diffuse ring:

• 0.0 – 0.3 °
• 0.3 – 0.5°
• 0.5 – 0.7 °
• 0.7 – 0.9 °
• 0.9 – 1.1 °
• 1.1 – 1.3 °
• 1.3 – 1.5 °

Offset Median:

• 0.22 °
• 0.41 °
• 0.6 °
• 0.8 °
• 1.0 °
• 1.2 °
• 1.4°

Associated FOV offset in the DL3: Median of the offset distribution in each ring

In Gammapy or Ctools, interpolation between the different offset

17

PSF

18

Methodology

• No theta2 applied
• Needed for 3D analysis
• PSF model:

→ list of MC containing true position and reconstructed position of each MC event → Build an histogram in Etrue and rad (distance from the true direction) weighted by the zenith and azimuth distribution of the events → fit by the sum of two Gaussians

• One test to evaluate our PSF model:

Distance from the true direction (degree)

100 – 200 GeV

PSF (sr-1)

→ 1D analysis: select a theta2 (θ0) region around the source to extract the signal

19

Correct the effective area

To be able to perform a spectral analysis (with a theta2 cut) from “full” IRF → need to correct the effective area by the PSF containment fraction

Corrected

Pointlike analysis phi1TeV = (4.08 +/- 0.29) 10-11 cm-2 TeV-1 s-1 Alpha= 2.54 +/- 0.09 Beta= 0.15 +/- 0.05 Full enclosure corrected analysis phi1TeV = (4.37 +/- 0.31) 10-11 cm-2 TeV-1 s-1 Alpha= 2.58 +/- 0.10 Beta= 0.18 +/- 0.05

Crab Nebula

20

Crab spectrum at different offset with Full IRF

phi1TeV = (3.74 +/- 0.1) 10-11 cm-2 TeV-1 s-1 Alpha= 2.48 +/- 0.03 Beta= 0.11 +/- 0.02 phi1TeV = (3.56 +/- 0.01) 10-11 cm-2 TeV-1 s-1 Alpha= 2.46 +/- 0.03 Beta= 0.08 +/- 0.01

Data set from 0.35, 0.4 and 0.7° from ST03-03 Full enclosure corrected Pointlike

21

Conclusions

• Pointlike and Full IRF
• Offset dependent IRF
• PSF

Perspectives

• pointlike DL3 release
• Produce Full DL3 release for the same sources
• PSF modelisation
• Background model?

22

Thanks for your attention