Approximating non-Gaussian Bayesian partitions with normalising flows: statistics, inference and application to cosmology
Tobias Röspel 1,⋆, Adrian Schlosser 1,⋆, Björn Malte Schäfer 1,2♯ · 2025
Read Approximating non-Gaussian Bayesian partitions with normalising flows: statistics, inference and application to cosmology by Tobias Röspel 1,⋆, Adrian Schlosser 1,⋆, Björn Malte Schäfer 1,2♯ (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-04791.txt.
Description
Tobias Röspel, Adrian Schlosser, and Björn Malte Schäfer use a machine learning trick to ease a heavy chore in statistics. Their 2025 paper concerns Bayesian inference for lopsided, non-Gaussian distributions, where sampling chains are slow and key summary values are hard to compute.
Their tool is the normalising flow, a map that bends a plain bell curve into the target shape and back. With it they compute partition functions, the normalizing values behind Bayesian posteriors, and extract means, spreads, and entropy more cheaply. They also find closed forms for posteriors past the Gaussian limit. A cosmology test case, fitting the matter density and a dark energy term to supernova data, shows the method at work.
Catalog record
| CREATOR | Tobias Röspel 1,⋆, Adrian Schlosser 1,⋆, Björn Malte Schäfer 1,2♯ |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-04791