Building Theoretical Models in a Data dominated Era
Lecturer: Oscar Macías (Virginia Tech).
Dates: August 3 - August 25, 2018
Place: 6-125. Tuesday-Thursday 10AM-12M
Registration: Send email to email@example.com
This set of five to six lectures will give an overview of the most common statistical tools used in astroparticle physics. Both theoretical concepts as well as computational methods will be covered: You will learn how parameters and parameter errors are estimated and how signal detections are quantified and reported. There will be hands-on tutorials with Python after the first lecture. Previous experience with python is beneficial but not required.
Given the limited time, the focus will be on the most common statistical concepts and methods for astroparticle physics.
Lecture 1: Introduction to data analysis, Poisson likelihood, Nuisance parameters, profile likelihood and Relation of likelihood to Bayesian statistics (prior and posterior).
Lecture 2: (Python Tutorial): Parameter estimation, confidence intervals, coverage, Detection (TS, Wilk’s theorem). The MINUIT algorithm.
Lecture 3: (Python Tutorial): Monte Carlo methods to estimate parameter errors, computation (and interpretation) of upper limits and sensitivity.
Lecture 4: (Python Tutorial): Frequentist and Bayesian computational methods: optimisation, sampling.
Lecture 5: Introduction to the GALPROP and DRAGON codes for numerical modelling.
Lecture 6: (Python tutorial): Propagate Dark Matter yields in the Galaxy with DRAGON/GALPROP and then test the simulations against available Cosmic Ray and/or gamma-ray data.