The Quantum Optimization Benchmarking Library

Area of Science:

• Quantum Computing
• Combinatorial Optimization
• Algorithm Benchmarking

Background:

• Benchmarking heuristic quantum algorithms at scale is becoming feasible.
• Empirical analysis is crucial for tracking progress towards quantum advantage in optimization.

Purpose of the Study:

• To introduce a systematic, fair, and comparable benchmarking framework for quantum optimization methods.
• To present ten model-independent problem classes challenging for classical methods.

Main Methods:

• Developed ten problem classes with varying properties, challenging classical methods with up to 100,000 decision variables.
• Created an open-source repository for track records of specific instances and solutions.
• Referenced state-of-the-art classical solver results and demonstrated exemplary quantum solver baselines.

Main Results:

• The framework enables standardized benchmark reporting for quantum optimization.
• Problem instances span a range of difficulty, from under 100 to 100,000 decision variables.
• Exemplary baseline results with quantum solvers are provided for selected problems.

Conclusions:

• The presented framework facilitates rigorous comparison of quantum and classical optimization algorithms.
• Standardized benchmarks are essential for driving the field towards demonstrable quantum advantage.
• The open-source repository supports reproducible research and community engagement.