Minion: Derivative-Free Optimization Library

Minion is a high-performance derivative-free optimization library designed for solving complex optimization problems where gradients are unavailable or unreliable. It features state-of-the-art evolutionary algorithms, including top-performing methods from IEEE CEC competitions, which are not commonly found in standard optimization libraries such as SciPy, NLopt, OptimLib, pyGMO, and pagmo2.

Minion also serves as a research platform for developing and testing new optimization algorithms. It includes benchmark functions from CEC competitions (2011, 2014, 2017, 2019, 2020, and 2022), providing a robust framework for algorithm evaluation and comparison.

🔥 Why Choose Minion?

• State-of-the-art optimization algorithms :
  • Differential Evolution-based algorithms:
    • Basic Differential Evolution (DE)
    • JADE
    • L-SHADE
    • jSO
    • j2020
    • NL-SHADE-RSP
    • LSRTDE
    • ARRDE (our novel Adaptive Restart-Refine DE algorithm)
  • Other population-based algorithms:
    • Artificial Bee Colony (ABC)
    • Grey Wolf DE Optimization
  • Classical optimization algorithms:
    • Nelder-Mead
    • Generalized Simulated Annealing (Dual Annealing)
    • L-BFGS-B (vectorized)
• Highly parallelized
  • Designed for vectorized function evaluations, supporting multithreading and multiprocessing to speed up optimization.
• Optimized C++ backend with Python API
  • Enjoy the performance of C++ with the simplicity of Python.
• CEC Benchmark Suite
  • Includes benchmark problems from 2011, 2014, 2017, 2019, 2020, and 2022 for rigorous algorithm testing.

🚀 Installation

The Python wrapper (minionpy) is available on PyPI:

pip install --upgrade minionpy

For compiling the C++ version, please refer to the official Minion documentation.

📖 Documentation

For full usage instructions, API reference, and examples, visit the official documentation:

• Minion Documentation