pyglobalsearch 0.4.0

This package provides a Python interface to Rust's globalsearch crate. globalsearch is a multistart framework for global optimization with scatter search and local NLP solvers written in Rust.

Python bindings for the globalsearch Rust crate

Website | Python API Docs | Examples

pyglobalsearch: Python interface for the Rust crate globalsearch. The Rust crate implements a modified version of the OQNLP (OptQuest/NLP) algorithm with the core ideas from "Scatter Search and Local NLP Solvers: A Multistart Framework for Global Optimization" by Ugray et al. (2007). It combines scatter search metaheuristics with local minimization for global optimization of nonlinear problems.

Similar to MATLAB's GlobalSearch [2]. The bindings are built using pyo3 and Maturin.

Installation

Install the Python package from PyPI:

pip install pyglobalsearch

Usage

1. Import the pyglobalsearch module:

   import pyglobalsearch as gs
   

2. Define the parameters of the optimization problem:

   params = gs.PyOQNLPParams(
    iterations=100,
    population_size=500,
    wait_cycle=10,
    threshold_factor=0.75,
    distance_factor=0.1,
   )
   

3. Define your problem:

   from numpy.typing import NDArray
   
   def objective(x: NDArray[np.float64]) -> float:
        return x[0] ** 2 + x[1] ** 2
   
   def gradient(x: NDArray[np.float64]) -> NDArray[np.float64]:
        return np.array([2 * x[0], 2 * x[1]])
   
   def variable_bounds() -> NDArray[np.float64]:
        return np.array([[-3, 3], [-2, 2]])
   
   problem = gs.PyProblem(objective, variable_bounds, gradient)
   

4. Run the optimization:

   result = gs.optimize(problem, params, local_solver="COBYLA", seed=0)
   print(result)
   

Developing

Prerequisites

• Rust
• uv
• Python (Python can also be installed using uv python install)
• Maturin

Setup

1. Clone the repository and navigate to the Python directory:

   git clone https://github.com/GermanHeim/globalsearch-rs.git
   cd globalsearch-rs/python
   

2. Create a virtual environment and activate it:

   uv venv
   source .venv/bin/activate
   

3. Modify the code and use Maturin to build the Python package:

   maturin develop
   

Limitations

The Python bindings are still in development and may not be fully functional. The API is subject to change. If you encounter any issues, please open an issue in the issue tracker or submit a pull request.

Additionally, the Python bindings have limitations:

• No support for checkpointing
• Maximum number of constraints is 1000

License

Distributed under the MIT License. See LICENSE.txt for more information.

References

[1] Zsolt Ugray, Leon Lasdon, John Plummer, Fred Glover, James Kelly, Rafael Martí, (2007) Scatter Search and Local NLP Solvers: A Multistart Framework for Global Optimization. INFORMS Journal on Computing 19(3):328-340. http://dx.doi.org/10.1287/ijoc.1060.0175

[2] GlobalSearch. The MathWorks, Inc. Available at: https://www.mathworks.com/help/gads/globalsearch.html (Accessed: 27 January 2025)