by Pranav Chandarana, Kipu Quantum, Germany

Abstract:

As combinatorial optimization increasingly shapes fields from logistics to machine learning, accelerating the search for high-quality solutions has become a central challenge. We show that bias-field digitized counterdiabatic quantum optimization (BF-DCQO) on IBM’s 156-qubit processors can already outperform leading classical solvers on difficult higher-order unconstrained binary optimization (HUBO) problems, achieving comparable or better solutions in seconds rather than minutes and demonstrating a clear, system-size-dependent runtime quantum advantage. To broaden and stabilize these gains, we introduce hybrid sequential quantum computing (HSQC), a flexible framework that interleaves classical heuristics, quantum optimization, and final classical refinement. Across representative implementations combining simulated annealing, BF-DCQO, and tabu-based or annealing-based post-processing, HSQC reliably recovers ground states within seconds and yields speedups of up to 700× over SA and 9× over MTS. These results highlight how purpose-built quantum algorithms integrated into structured hybrid workflows can deliver practical, scalable performance improvements on today’s quantum hardware.