The Python Surrogate Optimization Toolbox (pySOT) is an asynchronous parallel optimization toolbox for computationally expensive global optimization problems. pySOT is built on top of the Plumbing for Optimization with Asynchronous Parallelism (POAP), which is an event-driven framework for building and combining asynchronous optimization strategies. POAP has support for both threads and MPI.
pySOT implements many popular surrogate optimization algorithms such as the Stochastic RBF (SRBF) and DYCORS methods by Regis and Shoemaker, and the SOP method by Krityakierne et. al. We also support Expected Improvement (EI) and Lower Confidence Bounds (LCB), which are popular in Bayesian optimization. All optimization algorithms can be used in serial, synchronous parallel, and asynchronous parallel and we support both continuous and integer variables.
The toolbox is hosted on GitHub: https://github.com/dme65/pySOT
Documentation: http://pysot.readthedocs.io/
Installation instructions are available at: http://pysot.readthedocs.io/en/latest/quickstart.html
Several pySOT examples and notebooks can be found at:
https://github.com/dme65/pySOT/tree/master/examples
https://github.com/dme65/pySOT/tree/master/notebooks
If you use pySOT, please cite the following paper: David Eriksson, David Bindel, Christine A. Shoemaker. pySOT and POAP: An event-driven asynchronous framework for surrogate optimization. arXiv preprint arXiv:1908.00420, 2019
@article{eriksson2019pysot,
title={pySOT and POAP: An event-driven asynchronous framework for surrogate optimization},
author={Eriksson, David and Bindel, David and Shoemaker, Christine A},
journal={arXiv preprint arXiv:1908.00420},
year={2019}
}