Finding gamma-ray pulsars with sparse Bayes Blocks
Abstract
Beamed radiation from rapidly spinning (periods ~1-100ms), highly magnetized (~1012 gauss fields) neutron stars, or pulsars, is notoriously difficult to find in γ-rays. First, one may have to wait >10 minutes between photons even with a large gamma-ray telescope like CGRO-EGRET viewing a bright source like the Vela pulsar. Second, these γ-ray light-curves (brightness versus time or phase) are very sharply peaked. Current methods are carefully studied and well understood but use Fourier components - a bad match to this shape of light curve. Binning a light-curve into increasingly narrow bins then testing for flatness can introduce many free parameters and hence lower detection thresholds. So, why not use a statistic that more directly represents the sharp changes in a pulsar light curve? Why not let the data themselves (plus any prior knowledge) set the optimal size of a very few bins? This is what we have done. We test a modified "Bayesian Block" method on simulated light-curves with a variety of signal-to-noise-ratios. Preliminary results are encouraging, showing the "Sparse-BB" method more powerful for detecting very "spiky" light-curves.
- Publication:
-
Statistical Challenges in Astronomy
- Pub Date:
- 2003
- Bibcode:
- 2003sca..book..403C
- Keywords:
-
- Statistics;
- Gamma-Ray Pulsars;
- Bayesian Inference