EBL Absorption Signatures in DC2 Data

Jennifer Carson (SLAC)

DC2 closeout meeting

June 1, 2006

Motivation

• How well can we measure EBL absorption in 55 days of data?

• How sure can we be that we are distinguishing intrinsic spectral breaks from absorption signatures?

Fit spectra with broken power law model + absorption.

Compare results to simple power law fits.

Basic Procedure• v2 of catalog• 10° regions around LR’s 9 sources, 20° source regions• Only class A events• Diffuse components: extragalactic, galactic, high-energy

residual, low-energy residual• Diffuse prefactors allowed to float in fits• All catalog point sources in FOV included (fixed PLs)• Broken power law model with EBL absorption

(“BPLExpCutoff”): prefactor, 2 indices, break energy, Eabs, P1

= (E – Eabs) / P1

• Likelihood twice: DRMNGB for convergence, MINUIT for errors

Example Fits

bright blazar (BPL+EBL)

faint blazar(PL)

residual background

residual background

bright blazar (BPL+EBL)

galactic & EG

BPL Results

BPL KneiskeKneiske High-UVPrimackStecker

BPL Results

BPL PL

KneiskeKneiske High-UVPrimackStecker

BPL Results vs. PL Results

• PL E0 higher than BPL E0 for most sources

• Intrinsic breaks can pull E0 to lower energies

BP

L E

0 (

Ge

V)

PL E0 (GeV)

High-Redshift Sources• Found all sources with z > 2 in ASDC catalog (10)• ROI = 5°, source region = 15°

High-Redshift Sources• Found all sources with z > 2 in ASDC catalog (10)• ROI = 5°, source region = 15°

Can’t get good fits!(Despite hand-holding)

Fit Comparison: Idea

• Three types of fits:– Broken power law + fixed EBL absorption– Broken power law + no EBL absorption– Simple power law + floating EBL absorption

• Four free parameters in each fit

• Test statistic goodness of fit

Fit Comparison: Results

• red: TSBPL+EBL – TSBPL

• blue: TSBPL+EBL – TSPL+EBL

• TS generally better for BPL+EBL vs. pure BPL

• TS generally better for BPL+EBL vs. PL+EBL

Conclusions

• In < 2 months, difficult to constrain the EBL – no models ruled out

• PL & BPL results are close, but BPL generally predicts higher E0 values

• Especially difficult measure E0 for z > 2 sources• TS indicates that:

– BPL+EBL better fit than PL+EBL– BPL+EBL better fit than pure BPL

• High-energy residuals made finding E0 even harder• Future work

– More data! What would a year of data tell us?– Automated search for flares + spectral analysis– Improve cuts to remove high-energy residual.– Need likelihood to converge reliably with correct errors.