TEMPEST

Tempest is a Python implementation of the Persistent Sampling method for accelerated Bayesian inference

Getting started

Brief introduction

Tempest is a Python package for fast Bayesian posterior and model evidence estimation. It leverages the Persistent Sampling (PS) algorithm, offering significant speed improvements over traditional methods like MCMC and Nested Sampling. Ideal for large-scale scientific problems with expensive likelihood evaluations, non-linear correlations, and multimodality, Tempest provides efficient and scalable posterior sampling and model evidence estimation. Widely used in cosmology and astronomy, Tempest is user-friendly, flexible, and actively maintained.

Documentation

Read the docs at tempest-sampler.readthedocs.io for more information, examples and tutorials. For a detailed list of changes, see the CHANGELOG.md.

Installation

To install tempest using pip run:

pip install tempest-sampler

or, to install from source:

git clone https://github.com/minaskar/tempest.git
cd tempest
pip install .

Basic example

For instance, if you wanted to draw samples from a 10-dimensional Rosenbrock distribution with a uniform prior, you would do something like:

import tempest as tp
import numpy as np

n_dim = 10  # Number of dimensions

# Define prior transform: U(-10, 10) for each dimension
def prior_transform(u):
    return 20 * u - 10

# Define log-likelihood
def log_likelihood(x):
    return -np.sum(10.0*(x[:,::2]**2.0 - x[:,1::2])**2.0 \
            + (x[:,::2] - 1.0)**2.0, axis=1)

# Create and run sampler
sampler = tp.Sampler(
    prior_transform=prior_transform,
    log_likelihood=log_likelihood,
    n_dim=n_dim,
    vectorize=True,
)
sampler.run()

samples, weights, logl = sampler.posterior() # Weighted posterior samples

logz, logz_err = sampler.evidence() # Bayesian model evidence estimate and uncertainty

Attribution & Citation

Please cite the following papers if you found this code useful in your research:

@article{karamanis2025persistent,
  title={Persistent Sampling: Enhancing the Efficiency of Sequential Monte Carlo},
  author={Karamanis, Minas and Seljak, Uro{\v{s}}},
  journal={Statistics and Computing},
  volume={35},
  number={5},
  pages={1--22},
  year={2025},
  publisher={Springer}
}

Licence

Copyright 2026-Present Minas Karamanis and contributors.

Tempest is free software made available under the MIT License. For details see the LICENCE file.