Skip to Main content Skip to Navigation
Book sections

Clustering Milky Way's Globulars: a Bayesian Nonparametric Approach

Julyan Arbel 1
1 MISTIS - Modelling and Inference of Complex and Structured Stochastic Systems
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
Abstract : This chapter presents a Bayesian nonparametric approach to clustering , which is particularly relevant when the number of components in the clustering is unknown. The approach is illustrated with the Milky Way's glob-ulars, that are clouds of stars orbiting in our galaxy. Clustering globulars is key for better understanding the Milky Way's history. We define the Dirichlet process and illustrate some alternative definitions such as the Chinese restaurant process, the Pólya Urn, the Ewens sampling formula, the stick-breaking representation through some simple R code. The Dirichlet process mixture model is presented, as well as the R package BNPmix implementing Markov chain Monte Carlo sampling. Inference for the clustering is done with the variation of information loss function.
Complete list of metadatas

Cited literature [19 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01950656
Contributor : Julyan Arbel <>
Submitted on : Tuesday, December 11, 2018 - 3:32:15 AM
Last modification on : Thursday, March 26, 2020 - 8:49:33 PM
Document(s) archivé(s) le : Tuesday, March 12, 2019 - 12:31:13 PM

File

chapter_BNP.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01950656, version 1

Collections

Citation

Julyan Arbel. Clustering Milky Way's Globulars: a Bayesian Nonparametric Approach. Statistics for Astrophysics: Bayesian Methodology, pp.113-137, 2018. ⟨hal-01950656⟩

Share

Metrics

Record views

143

Files downloads

267