📝 SummarXiv

Abstract

We present QUBOLite, a Python package for the creation, manipulation, analysis, and solution of Quadratic Unconstrained Binary Optimization (QUBO) instances. Built as a thin wrapper around NumPy arrays, QUBOLite combines efficient numerical operations with high-level abstractions for tasks ranging from instance generation to preprocessing, analysis, and solving strategies, both exact and approximate. The package includes implementations of the QPRO+ algorithm by Glover et al. for identifying strong persistencies, dynamic range reduction heuristics, and an expressive system for partial assignments (clamping) enabling implicit variable assignment. The package is available on GitHub and the official Python package repository.