[Lisa_seminaires] NEW LOCATION: UdeM-McGill-MITACS machine learning seminar, PAA 1360, Fri Nov 30, 11:30am
Hugo Larochelle
larocheh at IRO.UMontreal.CA
Jeu 29 Nov 15:58:45 EST 2007
This week's seminar (see
http://www.iro.umontreal.ca/article.php3?id_article=107&lang=en):
A Bayesian approach to analyze water tank signals in the Pierre Auger
experiment
by Balazs Kegl,
Laboratoire de l’Accélérateur Linéaire
Université Paris-Sud 11
Location: Pavillon André-Aisenstadt (UdeM), room 1360
Time: November 30th 2007, 11h30
The objective of the Pierre Auger experiment (auger.org) is to study the
properties of ultra-high energy cosmic ray particles. When one if these
particles collides with the atmosphere, it generates a huge shower of
atmospheric particles that covers several square kilometers on the
Earth’s surface. The surface detector of the Auger Observatory (built on
the pampas of Argentina) consists of 1600 water tanks spaced at 1.5 km
on a regular hexagonal grid that detect atmospheric shower particles
through their interaction with water. To analyze the water tank signals,
we have built an elaborate generative model based on physical laws and
simulations of both the showers and the detector. To estimate the model
parameters, we use a Bayesian approach based on reversible jump Monte
Carlo Markov chains. This technique allows us to estimate the number of
muonic mixture components in the signal which is one of the best
indicators of the nature of the original cosmic ray particle. In the
first part of the talk we introduce the basic physics behind the
experiment and describe the statistical model. In the second part we
show the Bayesian technique that we are using to estimate the parameters
and present some preliminary results.
_______________________________________________
Lisa_seminaires mailing list
Lisa_seminaires at mercure.iro.umontreal.ca
https://webmail.iro.umontreal.ca/mailman/listinfo/lisa_seminaires
Plus d'informations sur la liste de diffusion Lisa_seminaires