SAMBO: Sequential And Model-Based (Bayesian) Optimization of black-box objective functions.

Project website

Documentation

Installation

$ pip install sambo
# or
$ pip install 'sambo[all]'   # Pulls in Matplotlib, scikit-learn

Usage

See examples on the project website.

Features

• Python 3+
• Simple usage, standard API.
• Algorithms prioritize to minimize number of evaluations of the objective function: SHGO, SCE-UA and SMBO available.
• Minimal dependencies: NumPy, SciPy (scikit-learn & Matplotlib optional).
• State-of-the-art performance—see benchmark results against other common optimizer implementations.
• Integral, real (floating), and categorical dimensions.
• Fast approximate global black-box optimization.
• Beautiful Matplotlib charts.

Development

Check CONTRIBUTING.md for hacking details.