IC22 Unbinned GRB Search Utrecht Collaboration Meeting Erik Strahler UW-Madison 16/9/2008
IC22 Data Sample 43/44 Northern hemisphere quality bursts Filtered to L3 Use complete year for background rejection Yields good statistics (~77M events) Bursts windows taken from the Swift T100s Conservative Easier than trying to figure out T90 information + padding Total ontime: 4961.3s Blind runs containing GRB triggers
Simulation Corsika Dataset: 645 (0.3 days) Corsika Coincident Dataset: 861 (0.71 days) Neutrino-Generator Datasets: 651 (E − 1, 1M) 768 (E − 2, 5.5M) Processed identically to data Weighted to average WB prompt emission spectrum Flux Normalization: 4.5E-9 GeV s-1 sr-1 cm-2 To Do: look at more muon statistics for checking tails
Cut Parameters
Bayesian Ratio Reduced LLH vs. Ndir Split Reco MinZen
L4 = L3 + paraboloid fit succeeded AND split reco succeeded AND neutrinos >= 90 degrees
“Loose” Cuts Pbf_status == 0 @ paraboloid error <5 Split reco minimum zenith angle >70 degrees Bayesian likelihood ratio > 20 32 iter. Pandel reduced log-likelihood < 8.25 Umbrella reduced log-likelihood < 15
Efficiency vs. Energy ~50% at 100 TeV (peak of WB signal acceptance)
Efficiency vs. NCH
Effective Area
Effective Area
Bayesian ratio > 20, rllh < 8.25
Likelihood Method Signal Spatial PDF signal Similar Method to GRB080319B except with 43 stacked sources Also incorporates energy as in the point source work Currently using fixed E-2 signal. (WB makes little change) Doesn’t make sense to fit the spectrum for 43 bursts and only 1 or 2 events Perform algorithm while optimizing rllh and bayesian ratio Signal Spatial PDF x PDFiE signal atm. corsika signal
Test Statistic 0 injected signal
Optimized Cuts 3s, P=0.5 MDP optimization 4s, P=0.5 MDP optimization Reduced LLH Bayesian LLH Ratio Reduced LLH Bayesian LLH Ratio
Discovery Potential Energy does not help as much as in the point source search ~15% improvement at P=0.5 for 4s
Conclusions Good event selection Similar to point source search Nearly at atm. neutrino level Good signal retention (0.7 events expected) Implemented likelihood function incorporating position, time, and energy (as nch) Initial tests show good discovery potential To Do: Data stabililty Run statistics for 5s numbers Implement individual GRB flux expectations