DATA ANALYSIS: EXTRACTING SCIENCE FROM MAGIC Uta Menzel YSW Ringberg 2015, 6. 7. 2015
OUTLINE MAGIC telescopes Imaging Air Cherenkov Technique (IACT) Telescope hardware Standard analysis Signal extraction Hillas parametrization Determination of physical parameters Final products My work - Template analysis Uta Menzel YSW Ringberg 2015, 6. 7. 2015 2 / 27
THE ATMOSPHERE Uta Menzel YSW Ringberg 2015, 6. 7. 2015 3 / 27
EXTENSIVE AIR SHOWERS Electromagnetic shower e + Primary e + e - e - e + e - e + e - e - e + particle cascade with pair production bremsstrahlung charged particles, faster than speed of light in air Cherenkov radiation Uta Menzel YSW Ringberg 2015, 6. 7. 2015 4 / 27
Data Analysis: Extracting science from MAGIC I MAGING A IR C HERENKOV T ECHNIQUE I image with elliptic shape I stereo for better reconstruction I background from cosmic rays, ~factor 1000 Christian Fruck Uta Menzel YSW Ringberg 2015, 6. 7. 2015 5 / 27
Data Analysis: Extracting science from MAGIC Main Imaging Air Cherenkov Telescopes Uta Menzel YSW Ringberg 2015, 6. 7. 2015 6 / 27
Data Analysis: Extracting science from MAGIC The MAGIC telescopes I 2200 m above sea level I parabolic dish: 17 m diameter I 2 telescopes, 85 m apart I focal length: 17 m I light structure fast repositioning I 1 m2 and 0.25 m2 mirrors Uta Menzel YSW Ringberg 2015, 6. 7. 2015 7 / 27
Data Analysis: Extracting science from MAGIC C AMERA Uta Menzel I PMTs (photo multiplier tubes) as photon detectors I 1039 pixels per camera I ~1 m diameter I 3.5 field of view YSW Ringberg 2015, 6. 7. 2015 8 / 27
ELECTRONICS trigger L0: minimum signal in single pixel L1: signal in neighbouring pixels L3: trigger in both telescopes readout: DRS4 chip up to 2 GSamples/s 1024 capacitors saved to disk: 60 time slices with charge (30 ns in total) for each pixel Uta Menzel YSW Ringberg 2015, 6. 7. 2015 9 / 27
DATA ANALYSIS Uta Menzel YSW Ringberg 2015, 6. 7. 2015 10 / 27
SIGNAL EXTRACTION pedestal subtraction search for signal (sliding window) for each pixel: number of photo electrons arrival time Uta Menzel YSW Ringberg 2015, 6. 7. 2015 11 / 27
IMAGE CLEANING Two-level image cleaning core pixel: number of photons above threshold 1 boundary pixel: neighbour of core pixel, number of photons above threshold 2 <threshold 1 include time contraints for core and boundary calibrated image core and boundary cleaned image Uta Menzel YSW Ringberg 2015, 6. 7. 2015 12 / 27
PARAMETRIZATION Mono Hillas Parameters number of photons length Source position distance length α center of gravity width width distance to source α time parameters... Christian Fruck Uta Menzel YSW Ringberg 2015, 6. 7. 2015 13 / 27
PARAMETRIZATION Stereo Hillas Parameters Reconstructed origin theta disp 2 Source position length 2 distance 1 distance 2 disp 1 α 2 length 1 center of gravity 2 α 1 width 2 number of photons length width distance to source α time parameters... Camera 1 Camera 2 center of gravity 1 width 1 Uta Menzel YSW Christian Ringberg Fruck 2015, 6. 7. 2015 13 / 27
RECONSTRUCTION based on Monte Carlo simulations atmosphere as part of the detector simulate the telescope response analyze like data parameters to reconstruct: energy of the primary particle: look-up table direction the of primary particle: stereo disp method hadronness: Random Forest Uta Menzel YSW Ringberg 2015, 6. 7. 2015 14 / 27
DIRECTION RECONSTRUCTION geometrical stereo reconstruction mono DISP reconstruction with Random Forest stereo DISP combine DISP of both telescopes to get direction Uta Menzel YSW Ringberg 2015, 6. 7. 2015 15 / 27
GAMMA-HADRON SEPARATION Electromagnetic shower Primary Hadron shower Cosmic Ray (p,, Fe...) e + e - Atmospheric Nucleus e + e - e + e - e + e - e - e + EM Shower EM Shower Nucleons, K, etc. Atmospheric Nucleus Nucleons, K, etc. e - + e - e + e+ EM Shower Uta Menzel YSW Ringberg 2015, 6. 7. 2015 16 / 27
GAMMA-HADRON SEPARATION Gamma Hadron Uta Menzel YSW Ringberg 2015, 6. 7. 2015 17 / 27
GAMMA-HADRON SEPARATION Decision tree Random Forest = many decision trees build with MC-gammas and real background determine hadronness of each event Uta Menzel YSW Ringberg 2015, 6. 7. 2015 18 / 27
SIGNAL DETECTION θ 2 -plot Skymap angular resolution ~0.1 background estimated from camera region without source Uta Menzel YSW Ringberg 2015, 6. 7. 2015 19 / 27
EFFECTIVE COLLECTION AREA A eff [m 2 ] determined from MCs (same analysis as data) A eff = N survived N simulated A simulated Uta Menzel YSW Ringberg 2015, 6. 7. 2015 20 / 27
MIGRATION MATRIX relation between simulated and reconstructed energy energy spectrum needs unfolding energy resolution ~15% Uta Menzel YSW Ringberg 2015, 6. 7. 2015 21 / 27
SPECTRUM Uta Menzel YSW Ringberg 2015, 6. 7. 2015 22 / 27
ALTERNATIVE ANALYSIS METHOD Uta Menzel YSW Ringberg 2015, 6. 7. 2015 23 / 27
Current analysis Calibrated data Template analysis Hillas method compare to template images Calibrated data Uta Menzel YSW Ringberg 2015, 6. 7. 2015 24 / 27
TEMPLATE DATABASE parameters that change the image: energy azimuth zenith angle impact parameter impact angle (ϕ) first interaction height 6D-database + pointing: (x cam, y cam ) source Uta Menzel YSW Ringberg 2015, 6. 7. 2015 25 / 27
TEMPLATE ANALYSIS PROCEDURE E, source pos. direction,... Hillas template database compare data with template new starting point data - Energy - Source position - Fit quality? comparison: likelihood function maximization of the likelihood Uta Menzel YSW Ringberg 2015, 6. 7. 2015 26 / 27
CONCLUSION IACT: new window to gamma-ray astronomy MAGIC: largest IACT stereo system standard analysis: based on image parameters possible improvement: template analysis better angular resolution better energy resolution higher sensitivity Uta Menzel YSW Ringberg 2015, 6. 7. 2015 27 / 27