Golem: An algorithm for robust experiment and process optimization

Golem is a Python tool that allows to compute the expectation and variance of black-box objective functions based on specified uncertainty/noise in the input variables. It can thus be used to see how different levels of input uncertainty might affect the location of the optimum, or it can be used in conjunction with optimization algorithms to enable robust optimization.

At the basis of the algorithm is the use of supervised tree-based models, such as regression trees and random forests. Please refer to the publication for the details of the approach.

User documentation

• Installation
• Golem Class
• Distributions
• Examples

Citation

If you use Golem in scientific publications, please cite the following paper:

• 13. Aldeghi, F. Häse, R.J. Hickman, I. Tamblyn, A. Aspuru-Guzik. Golem: An algorithm for robust experiment and process optimization. arXiv (2021), 2103.03716

@misc{golem,
  title={Golem: An algorithm for robust experiment and process optimization},
  author={Matteo Aldeghi and Florian Häse and Riley J. Hickman and Isaac Tamblyn and Alán Aspuru-Guzik},
  year={2021},
  eprint={2103.03716},
  archivePrefix={arXiv},
  primaryClass={math.OC}
  }

License

Golem is distributed under an MIT License.