New Zealand Statistical Association 2024 Conference

Kang Wang

Victoria University of Wellington

Galaxy cluster gas pressure profile modelling with reversible jump MCMC: A case study using Planck data

This is joint work with Yvette C. Perrott, Richard Arnold, David Huijser

This study introduces an innovative approach to modelling galaxy cluster gas profiles by combining Reversible Jump Markov Chain Monte Carlo (RJMCMC) with Nested Sampling. Traditional parametric methods, such as the generalised Navarro-Frenk-White (gNFW) profile, often face challenges like parameter degeneracy. In contrast, our method uses a flexible, semi-parametric nodal model to accurately define the gas pressure profile of galaxy clusters. This node-based model allows for automatic trans-dimensional model selection within a single program execution, eliminating the need to run multiple models and compare Bayes factors. Using data from the Coma, A2255, and A85 clusters observed by the Planck space telescope, our approach significantly improves the ability to describe the pressure-radius relationship compared to conventional parametric models.