sbi: Simulation-Based Inference

Getting Started | Documentation

sbi is a Python package for simulation-based inference, designed to meet the needs of both researchers and practitioners. Whether you need fine-grained control or an easy-to-use interface, sbi has you covered.

With sbi, you can perform simulation-based inference (SBI) using a Bayesian approach: Given a simulator that models a real-world process, SBI estimates the full posterior distribution over the simulator’s parameters based on observed data. This distribution indicates the most likely parameter values while additionally quantifying uncertainty and revealing potential interactions between parameters.

Key Features of sbi

sbi offers a blend of flexibility and ease of use:

• Low-Level Interfaces: For those who require maximum control over the inference process, sbi provides low-level interfaces that allow you to fine-tune many aspects of your workflow.
• High-Level Interfaces: If you prefer simplicity and efficiency, sbi also offers high-level interfaces that enable quick and easy implementation of complex inference tasks.

In addition, sbi supports a wide range of state-of-the-art inference algorithms (see below for a list of implemented methods):

• Amortized Methods: These methods enable the reuse of posterior estimators across multiple observations without the need to retrain.
• Sequential Methods: These methods focus on individual observations, optimizing the number of simulations required.

Beyond inference, sbi also provides:

• Validation Tools: Built-in methods to validate and verify the accuracy of your inferred posteriors.
• Plotting and Analysis Tools: Comprehensive functions for visualizing and analyzing results, helping you interpret the posterior distributions with ease.

Getting started with sbi is straightforward, requiring only a few lines of code:

from sbi.inference import NPE
# Given: parameters theta and corresponding simulations x
inference = NPE(prior=prior)
inference.append_simulations(theta, x).train()
posterior = inference.build_posterior()

Installation

sbi requires Python 3.9 or higher. While a GPU isn't necessary, it can improve performance in some cases. We recommend using a virtual environment with conda for an easy setup.

To install sbi, follow these steps:

1. Create a Conda Environment (if using Conda):

   conda create -n sbi_env python=3.9 && conda activate sbi_env

2. Install sbi: Independent of whether you are using conda or not, sbi can be installed using pip:

pip install sbi

3. Test the installation: Open a Python prompt and run

from sbi.examples.minimal import simple
posterior = simple()
print(posterior)

Tutorials

If you're new to sbi, we recommend starting with our Getting Started tutorial.

You can also access and run these tutorials directly in your browser by opening Codespace. To do so, click the green “Code” button on the GitHub repository and select “Open with Codespaces.” This provides a fully functional environment where you can explore sbi through Jupyter notebooks.

Inference Algorithms

The following inference algorithms are currently available. You can find instructions on how to run each of these methods here.

Neural Posterior Estimation: amortized (NPE) and sequential (SNPE)

• (S)NPE_A (including amortized single-round NPE) from Papamakarios G and Murray I Fast ε-free Inference of Simulation Models with Bayesian Conditional Density Estimation (NeurIPS 2016).

• (S)NPE_C or APT from Greenberg D, Nonnenmacher M, and Macke J Automatic Posterior Transformation for likelihood-free inference (ICML 2019).

• TSNPE from Deistler M, Goncalves P, and Macke J Truncated proposals for scalable and hassle-free simulation-based inference (NeurIPS 2022).

• FMPE from Wildberger, J., Dax, M., Buchholz, S., Green, S., Macke, J. H., & Schölkopf, B. Flow matching for scalable simulation-based inference. (NeurIPS 2023).

• NPSE from Geffner, T., Papamakarios, G., & Mnih, A. Compositional score modeling for simulation-based inference. (ICML 2023)

Neural Likelihood Estimation: amortized (NLE) and sequential (SNLE)

• (S)NLE or just SNL from Papamakarios G, Sterrat DC and Murray I Sequential Neural Likelihood (AISTATS 2019).

Neural Ratio Estimation: amortized (NRE) and sequential (SNRE)

• (S)NRE_A or AALR from Hermans J, Begy V, and Louppe G. Likelihood-free Inference with Amortized Approximate Likelihood Ratios (ICML 2020).

• (S)NRE_B or SRE from Durkan C, Murray I, and Papamakarios G. On Contrastive Learning for Likelihood-free Inference (ICML 2020).

• (S)NRE_C or NRE-C from Miller BK, Weniger C, Forré P. Contrastive Neural Ratio Estimation (NeurIPS 2022).

• BNRE from Delaunoy A, Hermans J, Rozet F, Wehenkel A, and Louppe G. Towards Reliable Simulation-Based Inference with Balanced Neural Ratio Estimation (NeurIPS 2022).

Neural Variational Inference, amortized (NVI) and sequential (SNVI)

• SNVI from Glöckler M, Deistler M, Macke J, Variational methods for simulation-based inference (ICLR 2022).

Mixed Neural Likelihood Estimation (MNLE)

• MNLE from Boelts J, Lueckmann JM, Gao R, Macke J, Flexible and efficient simulation-based inference for models of decision-making (eLife 2022).

Feedback and Contributions

We welcome any feedback on how sbi is working for your inference problems (see Discussions) and are happy to receive bug reports, pull requests, and other feedback (see contribute). We wish to maintain a positive community; please read our Code of Conduct.

Acknowledgments

sbi is the successor (using PyTorch) of the delfi package. It started as a fork of Conor M. Durkan's lfi. sbi runs as a community project. See also credits.

Support

sbi has been supported by the German Federal Ministry of Education and Research (BMBF) through project ADIMEM (FKZ 01IS18052 A-D), project SiMaLeSAM (FKZ 01IS21055A) and the Tübingen AI Center (FKZ 01IS18039A). Since 2024, sbi is supported by the appliedAI Institute for Europe.

License

Apache License Version 2.0 (Apache-2.0)

Citation

If you use sbi consider citing the sbi software paper, in addition to the original research articles describing the specific sbi-algorithm(s) you are using.

@article{tejero-cantero2020sbi,
  doi = {10.21105/joss.02505},
  url = {https://doi.org/10.21105/joss.02505},
  year = {2020},
  publisher = {The Open Journal},
  volume = {5},
  number = {52},
  pages = {2505},
  author = {Alvaro Tejero-Cantero and Jan Boelts and Michael Deistler and Jan-Matthis Lueckmann and Conor Durkan and Pedro J. Gonçalves and David S. Greenberg and Jakob H. Macke},
  title = {sbi: A toolkit for simulation-based inference},
  journal = {Journal of Open Source Software}
}

The above citation refers to the original version of the sbi project and has a persistent DOI. Additionally, new releases of sbi are citable via Zenodo, where we create a new DOI for every release.