determining spatial extendedness of glast sources
DESCRIPTION
Determining Spatial Extendedness of GLAST Sources. Adam Zok Science Undergraduate Laboratory Internship Program August 14, 2008. GLAST: Key Concepts. High energy: 30 MeV – 300 G eV Limited spatial resolution: 0.15° - 3.5° Resolution worsens at low photon energies - PowerPoint PPT PresentationTRANSCRIPT
Determining Spatial Extendedness of GLAST Sources
Adam ZokScience Undergraduate Laboratory Internship Program
August 14, 2008
GLAST: Key ConceptsHigh energy: 30 MeV – 300 GeVLimited spatial resolution: 0.15° - 3.5°Resolution worsens at low photon energies
Coulomb scattering from heavy nucleiTargets of study: typically < 1°
Identifying SourcesMany potential gamma-ray emitters may lie
within GLAST’s spatial uncertaintySome emitters are not point sources, but are
spatially extended (they have a measurable angular size)
Spatially extended sources are much less common than point sources, so identifying one can narrow down the list of candidate objects significantly.
Software Toolsgtobssim
Creates virtual gamma-ray emitters, outputs .fits file that represents how GLAST may view the source
sourcefitWorks backwards: subtracts background radiation,
reconstructs source parameters and calculates confidence limits
Optimizes likelihood (probability that a given set of data came from a particular distribution)
Python modules: PyFITS, ROOT
Gtobssim Simulation
Testing Sourcefit OptionsSourcefit allows the user to specify certain
fitting options, or simply use the defaultsIn particular, I wanted to see how the energy
binning and energy range used affected fit quality
To determine how to most effectively use the program, I ran fits on the same sources using several different combinations of settings
Energy Ranges
Red: default range
Brown: 100 MeV – 100 GeV
Green: 500 MeV – 100 GeV
Blue: 1 GeV – 100 GeV
Energy Binning
Red: default binning (irregular)
Brown: 2 bins per decade
Green: 3 bins per decade
Blue: 4 bins per decade
Pink: 6 bins per decade
Determining Sourcefit’s LimitsNeeded to find out which kinds of sources could be
accurately modeled by sourcefitUsed two different fitting algorithms: Minuit and SimplexGenerated 4 arrays of simulated sources obscured by
background radiationDifferent flux for each array, varied size and spectrum
within the arraysInvestigated accuracy of fits in terms of size and position,
as well as the calculated confidence limits
Array Fit ResultsMinuit and Simplex performed comparablyBoth algorithms did a poor job of calculating reasonable
confidence limitsSources with a high flux and low spectral index (lots of
energetic photons) were most successfully parameterized for both size and position
Simplex Position Fitting ResultsFlux = 3 x 10-5 s-1m-2
Simplex Position Fitting ResultsFlux = 10-4 s-1m-2
Simplex Position Fitting ResultsFlux = 3 x 10-4 s-1m-2
Simplex Position Fitting ResultsFlux = 10-3 s-1m-2
TS Values, Flux = 10-3 s-1m-2
Minuit Point Source Fitting
Red = unacceptable fit ( > 0.01° )
Blue = good fit ( < 0.01° )
Green = very good fit ( < o.oo1° )
Final ThoughtsDefault energy range usually works best, but low flux,
soft spectrum sources may be better fit with a wider energy range (including more low energy photons)
TS value correlates most strongly with source size and spectral indexMore background (incorrectly) detected for large, soft-
spectrum sources
Future WorkProblems with error matrix calculated by sourcefit need
to be fixedArray plots that quantify error, instead of “yes” or “no”
classificationAnalyze sources with less regular spectraIntroduce background radiation from galactic sourcesAdditional simulations to rule out statistical irregularities