Speaker
Description
Many astroparticle experiments measure the same events with complementary detector systems. Calibrating one energy measurement against another is complicated by fluctuations in both measurements and by selection effects. This talk presents a hierarchical Bayesian framework that infers the true energy of each event jointly with the calibration function. The likelihood includes the detector responses, the distribution of true energies and the event selection. Neglecting any of these can bias the inferred calibration. The framework also lets the data determine how much structure in the calibration function can be resolved. Tests on synthetic data show unbiased recovery of the calibration function and correct posterior coverage. Developed for hybrid cosmic-ray measurements, the framework can be adapted to other experiments with complementary detectors and extended to include latent event properties such as composition.