München 2009 – scientific programme
Parts | Days | Selection | Search | Downloads | Help
T: Fachverband Teilchenphysik
T 90: Neutrino-Astronomie 3
T 90.4: Talk
Wednesday, March 11, 2009, 17:30–17:45, A240
Markov-Chain Monte Carlo Reconstruction for Cascades in IceCube — •Jakob van Santen1 and Eike Middell2 — 1Humboldt Universität zu Berlin, Fachbereich Physik, D-12489 Berlin, Germany — 2DESY Zeuthen, Platanenallee 6, D-15738 Zeuthen, Germany
In particle detector experiments, it is often necessary to reconstruct information about the incoming particle based on the detector response. One technique is to describe the likelihood of a certain detector response given an event hypothesis and then vary the hypothesis to maximize the likelihood.
Markov-Chain Monte Carlo (MCMC) techniques offer the ability to efficiently sample such a likelihood function in the most significant regions of a large parameter space. The MCMC generates a set of points in parameter space whose distribution is proportional to the likelihood function. The characteristics of this distribution can be used to judge the quality of a reconstruction and filter out poorly-reconstructed events.
I will discuss the application of MCMC techniques to the reconstruction of neutrino-induced cascade events in the IceCube neutrino detector.