guofei9987/scikit-opt

# Technical Summary Provides a pluggable operator framework where users can register custom selection, crossover, mutation, and ranking functions via `register()` method, enabling algorithmic experimentation beyond built-in implementations. Supports iterative continuation (resuming from previous runs), vectorization, multithreading, multiprocessing, and caching for function evaluation acceleration, with nascent GPU computation support. Targets general-purpose constrained and unconstrained optimization across continuous and discrete domains, integrating seamlessly with NumPy-based workflows.

6,388 stars. No commits in the last 6 months. Available on PyPI.