EBL attenuation of DC2 Sources Luis C. Reyes U. of Maryland / GSFC lreyes@milkyway.gsfc.nasa.gov DC2 Closeout May 31st, 2006
Introduction What is it? Extragalactic Background Light (EBL) is the build up of all forms of radiation emitted in the universe since the epoch of recombination (by convention it doesn't include the CMB) It contains informations about the history of star formation, large scale structure of the universe, epoch of recombination, etc Why is it important for GLAST science? Because extragalactic high energy gamma rays (E > few GeV) are absorbed by the EBL through pair production, therefore: In order to obtain the actual spectrum emitted by the source you need to account for EBL attenuation. Gamma ray observations can probe the density of the EBL by measuring the strength of the attenuation GLAST is sensitive to optical UV EBL, not infrared.
DC2 Analysis Objectives: Look for EBL attenuation of blazars Plausible for DC2 given bright and hard sources Compare results to the predictions from EBL models Caveat : 55 days is a short length of time in order to look for EBL attenuation effects: In principle, most sources experience the cutoff at energies ~ > 50GeV, where the fluxes are low. More integration time is desirable to get the statistics required for meaningful results.
Characterizing the Optical Depth ( τ ) The optical depth is a non trivial function of γ ray energy and redshift: comoving number density of EBL photons with energy ε at redshift z. From an observational point of view, the relevant range for optical depth is: ~0.01 < τ < ~4 ~1 < e τ < ~0.01 the parametrization to be used should reproduce well enough the optical depth in this range
Characterizing the Optical Depth ( τ ) EBL attenuation is sometimes characterized with F obs = F 0 e E/λ e E/λ Really problematic when fitting over several orders of magnitude If the optical depth is such that τ ~ a ln (E) for E>E b then e τ ~ E a and for E>E b flux ~ E α e τ = E α a EBL attenuation could mimic spectral breaks in the spectrum With one more degree of freedom, it seems we can characterize correctly the optical depth: Fits well over 5 orders of magnitude τ (E) = (E E b ) / P 1 ; if E > Eb With a very high number of statistics one could add more degrees of freedom to characterize the opacity more accurately Nevertheless, at the end it is an approximation!
The Source Model ExpCutoff is used to model the point sources (blazars): dn de = P x E ; if E E b S E exp E E b / P 1 ; if E E b S Parameters: P Prefactor (free) S Scale factor (fixed) α Index (free) E b Break energy (GeV) P 1 Attenuation Scale The unattenuated spectrum of the blazar is assumed to be a simple power law It is simple and sensitive to steepening of the spectrum of dim sources. Galactic and extragalactic diffuse are considered Contamination by residual background is considered following the analysis by Riccardo Rando.
Residual Background Contamination Seth studied the spectral and spatial distribution of residual background in DC2 data, he found that this contribution is comparable to the extragalactic diffuse at low and high energies. Bill Atwood implemented additional cuts that improve the BG rejection substantially at high energies (see link below). These cuts, however, required new variables, that were not implemented in time for the original DC2 data generation. The reprocessed DC2 data doesn't have these cuts either. Residual BG Galactic Diff Extragalactic Diff Riccardo Rando produced a MapCube that accounts for the residual background contribution. Region of the sky without point sources MapCube Spectrum
Results from Likelihood Source MRF0260 from Catalog v1 EG Diffuse Point Source Point Source: Prefactor = 66.94 ± 15.35 Index = 2.56 ± 0.08 Eb = 8.65 ± 2.72 P1 = 10.15 ± 5.54 Extragalactic Diffuse: Prefactor = 2.39 ± 2.15 Index = 2.1 (fixed) Galactic Diffuse: Value = (0.42 ± 1.5) x GALPROP Gal Diffuse Residual BG Residual BG: Value = (0.64 ± 0.38) x Riccardo's MapCube
Results from Likelihood Source Names from version 1 of LAT Catalog MRF0224 z=1.5 MRF0244 z=0.9 MRF0260 z=1.4 MRF0294 z=1.0 MRF0110 z=1.21 MRF0125 z=1.5 MRF0300 z=2.1 MRF0309 z=2.85 MRF0311 z=4.0
Results from Likelihood From the likelihood fit we obtain the parametrization of τ in terms of E b and P 1 We can calculate at which energy does τ(e o ) = 1? Interpretation *Source names from version 1 of LAT Catalog The likelihood finds sources with steepening, is it due to EBL attenuation?, is it a feature of the intrinsic spectrum of the source?, or both? No sensible answer can be given without any simultaneous observation at a different wavelength (which can be used to predict the emitted spectrum in γ rays, assuming a blazar emission model), or By observing a cutoff at the same energy for other sources with the same redshift! (if enough sources)
The Fazio Stecker Relation We don't get enough sources with meaningful results to verify that sources at the same redshift have a rolloff at the same energy. However, when comparing with the EBL models, the data points seem lo lie in a similar pattern The data supports a high opacity model
Sanity Check? Pre DC2 1 year long simulation Same method Blazar definition from Checkout3 blazars
Conclusions from the Analysis The DC2 data shows hints of EBL attenuation: Steepening in the spectra of bright, hard blazars. The rolloff energy (Eo) as function of redshift has a pattern that is consistent with EBL models No high redshift sources were found with high Eo No low redshift sources were found with low Eo There are a some effects that might distort the results: Intrinsic spectral features in the sources like energy breaks Likely, the residual background contamination is not perfectly accounted for. Something interesting to check after the truth is unveiled.
The DC2 Experience Science Tools (specially Likelihood) are powerful and work well once you get to know the peculiarities (optimizers, pylikelihood commands, etc.,...) DC2 sky was complex and challenging A few suggestions: To offer the possibility to have user defined source model types (like expcutoff) without having to compile code. Calculate 2 / ndf after the fit within likelihood Access to the optimizer information (error matrix in minuit for example) from gtpython
BACKUP SLIDES
How does the EBL affect γ ray sources? e 1 Model by Kneiske et al (2004) ~200 GeV For a given source at redshift z: F observed = e τ(e,z) F emitted ~30 GeV τ 0 for E < ~GeV for all redshifts γ rays with E > 300 GeV are strongly absorbed when emitted at z > ~ 0.5 GLAST will have the unique opportunity to measure the EBL attenuation of thousands sources from z > 1 up to z = ~4 (blazars) or z=6? (GRBs) With different approaches, several models predict the opacity due to the EBL. They are available for simulation in the GLAST software. For information see: https://confluence.slac.stanford.edu/display/scigrps/e BL+attenuation+of+Blazars+ +Simulation Lines in the plot satisfy τ(e,z) = 1
... You can create your own functions to be used with Likelihood Suggestion 1: It will be good to offer the user the possibility of defining these sources without having to compile code.