IBM and Kipu Quantum have published new experimental results showing that a hybrid quantum algorithm—Branch-and-Bound Digitized Counterdiabatic Quantum Optimization (BBB-DCQO)—can outperform both classical and quantum annealing methods in solving higher-order unconstrained binary optimization (HUBO) problems. The algorithm, tailored specifically to IBM’s superconducting hardware and the HUBO problem class, was run on the IBM Heron quantum processor, achieving solution runtimes in fractions of a second—well below the minutes required by standard commercial solvers.
The results, presented in two recent preprints, show that BBB-DCQO yields higher-quality solutions while requiring up to 10x fewer function evaluations compared to simulated and quantum annealing techniques. The Heron-based implementation solved benchmarked 100-qubit HUBO instances without converting them to the more common QUBO form, thus avoiding mapping inefficiencies. The hybrid method combines bias-field digitized counterdiabatic evolution with a branch-and-bound strategy, allowing for more targeted and efficient exploration of large solution spaces.
This milestone reflects a significant advance in near-term quantum optimization. IBM’s Jay Gambetta highlighted the importance of intelligent problem formulation in leveraging today’s gate-based systems for performance gains, emphasizing that algorithm-hardware co-design is now a critical factor alongside raw qubit scale. Kipu Quantum CEO Daniel Volz noted that the next challenge is surpassing state-of-the-art commercial solvers for real-world applications in logistics, finance, and energy.
Read more from IBM’s LinkedIn post here and access the full preprint on arXiv here.
May 16, 2025
