.rst .pdf

repository open issue suggest edit

Contents

Introduction

arianna 1 is a Python implementation of the Metropolis–Coupled Slice Sampling method that generates posterior samples from high-dimensional and strongly multimodal distributions. Apart from Bayesian parameter inference, arianna also provides unbiased and low-variance estimates of the model evidence (aka marginal likelihood) at no additional cost. The sampler is modular and does not require any hand-tuning from the user. Its parallel and black-box nature renders it ideal for computationally expensive models with high number of parameters often met in the physical sciences.

1

Named after Dr. Arianna W. Rosenbluth, one of the main developers of the Metropolis algorithm and the first person in history to ever code a Markov Chain Monte Carlo algorithm.

Documentation

Contents:

• Installation
  • Dependencies
  • Using pip
  • From source
• Quickstart guide
• API Reference
  • The Ensemble Sampler
  • The Chain Manager

Attribution

Please cite the following if you find this code useful in your research. The BibTeX entry for the paper is:

@article{arianna,
    title={arianna: A Metropolis--Coupled Slice Sampler for Bayesian Inference and Model Selection},
    author={Minas Karamanis and Florian Beutler},
    year={2021},
    note={in prep}
}

Authors & License

Copyright 2021 Minas Karamanis and contributors.

arianna is free software made available under the GPL-3.0 License.

Changelog

0.0.1 (14/03/21)

• First version

Installation