https://github.com/tillahoffmann/summaries
Comparison of summary statistic selection methods with a unifying perspective.
https://github.com/tillahoffmann/summaries
approximate-bayesian-computation conditional-density-estimation data-compression information-theory likelihood-free-inference simulation-based-inference summary-statistics
Last synced: about 1 year ago
JSON representation
Comparison of summary statistic selection methods with a unifying perspective.
- Host: GitHub
- URL: https://github.com/tillahoffmann/summaries
- Owner: tillahoffmann
- License: bsd-3-clause
- Created: 2022-01-14T21:07:26.000Z (about 4 years ago)
- Default Branch: main
- Last Pushed: 2023-06-23T20:00:24.000Z (over 2 years ago)
- Last Synced: 2025-01-14T02:46:38.782Z (about 1 year ago)
- Topics: approximate-bayesian-computation, conditional-density-estimation, data-compression, information-theory, likelihood-free-inference, simulation-based-inference, summary-statistics
- Language: Python
- Homepage: https://arxiv.org/abs/2206.02340
- Size: 229 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.rst
- License: LICENSE
- Citation: CITATION.cff
Awesome Lists containing this project
README
summaries
=========
.. image:: https://github.com/tillahoffmann/summaries/actions/workflows/main.yml/badge.svg
:target: https://github.com/tillahoffmann/summaries/actions/workflows/main.yml
This repository accompanies the preprint `"Minimizing the expected posterior entropy yields optimal summary statistics" `_ and can be used to reproduce all figures and reported results.
You can reproduce the results in three steps:
1. Set up a new, clean python virtual environment (you can also skip this step to use your host python environment, but your mileage may vary).
2. Install all the requirements by running :code:`pip install -r requirements.txt`.
3. Generate all result files by running the following code from the command line.
.. code-block:: bash
# Ignore the figures until we have generated all data files.
doit ignore figures
# Generate the data files (use -n to parallelize if desired).
doit -n [number of cores]
# Generate the figures and a summary file `workspace/figures/figures.html`.
doit forget figures
doit figures
You will find all figures in the folder :code:`workspace/figures` together with a HTML report :code:`workspace/figures/figures.html` that contains additional information. This process takes about 40 minutes on an M1 MacBook Pro when parallelizing across six cores.
The code has been tested with python 3.9 on macOS 12.4 (Monterey) running on Apple silicon and on Ubuntu 20.04. The :code:`summaries` package has complete test coverage, and you can run :code:`pytest` from the repository root to verify that your installation is working.