34th Annual Student Research Conference
Bayesian Inference and Cosmological Data Analysis
Andrew D. Gentry
Prof. Timothy Wiser, Faculty Mentor
Bayesian inference is a statistical method that is becoming increasingly popular for use in astrophysical and cosmological parameter estimation and model comparison. The key advantage of Bayesian inference is the ability to account for prior information when analyzing statistical error. Bayes’ theorem provides a quantitative formula for updating a hypothesis based on experiment. Bayesian inference is more useful than traditional “frequentist” methods when complicated cosmological theories combined with large, noisy datasets make it difficult to create a consistent way to analyze data. We present the statistical framework of Bayesian inference, especially focusing on its use in parameter estimation. Bayesian inference often necessitates the use of computational techniques. We discuss Markov Chain Monte Carlo (MCMC), the primary numerical tool used in modern cosmology. Several of the most popular MCMC algorithms will be reviewed, and an illustrative example using data from the cosmic microwave background will be presented.
Keywords: cosmology, bayes , bayes theorem, cosmic microwave background, Monte Carlo
Topic(s):Physics
Astronomy
Statistics
Presentation Type: Face-to-Face Oral Presentation
Session: 401-2
Location: SUB GEO
Time: 3:45