A Pythonic implementation of the Nested Sampling integration algorithm for Bayesian model comparison and parameter estimation.
This package provides multiple implementations for constrained drawing functions, and a test suite to evaluate the correctness, accuracy and efficiency of various implementations.
The source code consists of three components:
- A Modular framework for nested sampling algorithms (nested_sampling) and their development
- Test framework to evaluate the performance and accuracy of algorithms (testsuite)
- UltraNest - A fast C implementation of a mixed RadFriends/MCMC nested sampling algorithm
For drawing uniformly above the likelihood constraint, currently the following algorithms are implemented:
- naive rejection sampling
- MCMC with adaptive proposal
- RadFriends/SupFriends (see Buchner, 2014)
- SVM-based constrainer
- Galilean (in progress)
- Hamiltonian (in progress)
For testing the correctness and efficiency of algorithms, multiple algorithms can be run against a set of test problems.
Additionally to nested sampling implemented here, the following algorithms are evaluated in the benchmark:
- MultiNest (through PyMultiNest)
- Cuba-based algorithms (through PyCuba)
- Suave
- Divonne (through PyCuba)
- Cuhre (through PyCuba)
- Vegas (through PyCuba)
The evaluation of various algorithms can be found at https://johannesbuchner.github.io/UltraNest/testsuite/
UltraNest is a fast C implementation of a mixed RadFriends/MCMC nested sampling algorithm. It is documented at https://johannesbuchner.github.io/UltraNest/ultranest/ You are welcome to dive into the code (ultranest folder) and implement your preferred algorithm.
Have a look at the file nested_sampling/test/simplenested_test.py to build your own combination of nested sampling algorithm components.
You can run the entire test suite using the testsuite/testbase.py but to only run a part of it create skip_algorithms and skip_problems files and run testsuite/partialtestbase.py
The documentation is available at https://johannesbuchner.github.io/UltraNest/ The source code repository is available at https://github.com/JohannesBuchner/UltraNest/
The correct citation to use is Buchner (2014) published in Statistics and Computing (Journal article, Preprint PDF).
BibTeX:
@Article{Buchner2014stats,
Title = {{A statistical test for Nested Sampling algorithms}},
Author = {Buchner, Johannes},
Year = {2014},
Doi = {10.1007/s11222-014-9512-y},
Eprint = {1407.5459},
ISSN = {0960-3174},
Language = {English},
Month = jul,
Pages = {1-10},
Adsnote = {Provided by the SAO/NASA Astrophysics Data System},
Adsurl = {http://adsabs.harvard.edu/abs/2014arXiv1407.5459B},
Archiveprefix = {arXiv},
Journal = {Statistics and Computing},
Keywords = {Nested sampling; MCMC; Bayesian inference; Evidence; Test; Marginal likelihood},
Owner = {user},
Primaryclass = {stat.CO},
Publisher = {Springer US},
Timestamp = {2014.08.20}
}
This work is open source software licensed under GPLv3 Affero (see LICENSE file). If you are interested in a different license, please contact me.
