← Latest papers
⚛️ quantum physics

Proven optimality for LABS with quantum-classical branch and bound

This paper introduces a quantum-classical branch and bound algorithm that leverages fixed-angle QAOA sampling to solve the Low Autocorrelation Binary Sequence (LABS) optimization problem, successfully proving global optimality for sequence lengths up to 30 while providing certified lower bounds and convergence metrics without relying on quantum speedup.

Original author: András Czégel

Published 2026-11-20
📖 5 min read🧠 Deep dive

Original author: András Czégel

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer

In the hidden architecture of modern technology, there exists a class of mathematical puzzles known as low autocorrelation binary sequences. These are not abstract curiosities but the fundamental building blocks of how we communicate and sense the world. When a radar system sends out a pulse to detect an aircraft, or when a digital signal synchronizes a mobile phone connection, it relies on a specific pattern of positive and negative values. The quality of this pattern determines whether the system sees a clear target or a confusing mess of false echoes. The goal is to find a sequence where the internal overlaps, or correlations, are as small as possible. This is a notoriously difficult problem, akin to searching for a single, perfectly smooth needle in a haystack of billions of jagged ones. For decades, computers have struggled to prove they have found the absolute best pattern, often settling for very good guesses without knowing if a better one exists.

A researcher named András Czégel has now demonstrated a way to solve this problem with absolute certainty, using a hybrid approach that blends classical computing with quantum sampling. The work does not claim to be faster than the best traditional methods; in fact, the author explicitly states there is no speedup. Instead, the achievement lies in the proof. By embedding a quantum algorithm inside a rigorous classical search framework, the team proved the optimal solutions for sequences ranging from length 2 to length 30. This is a significant milestone because previous quantum attempts on this problem could only find good sequences but could never guarantee they were the best, nor could they measure how far a guess might be from perfection. This new method provides a certificate of optimality, a mathematical guarantee that no better sequence exists, along with a precise measurement of how the search converged over time.

The core of the method is a "branch and bound" strategy, a systematic way of exploring possibilities by cutting off dead ends. Imagine a tree growing in every direction, where each branch represents a decision to set a value in the sequence. A classical computer usually walks this tree, checking bounds to see if a branch can possibly lead to a better solution. If a branch looks too expensive, it is pruned away. Czégel's innovation was to place a quantum sampler, specifically a fixed-angle quantum algorithm, inside this classical tree. At every node where a decision must be made, the quantum sampler takes a quick look at the landscape of possibilities. It does not solve the whole problem; it simply offers a candidate for the best solution found so far and suggests which path to explore next. The classical computer then uses these suggestions to guide the search, while its own strict mathematical rules prune the tree and verify the final result.

The results of this hybrid approach were measured on sequences up to length 30. For every length in this range, the algorithm successfully proved the optimal energy value, meaning it confirmed that no other sequence could have a lower sum of squared correlations. The process was tested with different settings, including a depth of 12 for the quantum sampler and a shallower depth of 1. In the deeper setting, the quantum sampler found the best possible solution immediately for many cases, allowing the classical part of the system to focus on proving that no better solution existed. For longer sequences, specifically those of length 27 and above, the researchers found that the main problem was too large for the quantum sampler to handle directly, so they split the problem into smaller, manageable pieces based on symmetry before sampling. Even in these complex cases, the system returned a certified proof of optimality.

The study also carefully examined what the quantum sampler actually contributed to the success. It turned out that the sampler's primary value was in finding the best solution early in the process. Once the best solution was found, the quantum sampler's suggestions for which path to take next did not always improve the search; in some cases, a simple random choice or a fixed rule performed just as well or even better. This suggests that the quantum sampler is not a magic wand that instantly solves the puzzle, but rather a tool that helps the classical computer find the target faster, allowing the rigorous proof machinery to take over. The researchers ran simulations on noiseless computers and also modeled how the system would behave on real quantum hardware, such as devices from IBM and Quantinuum. They found that while current hardware is not yet powerful enough to handle the largest problems with high fidelity, the method remains valid even with noise, because the classical part of the algorithm provides the final proof of correctness.

This work represents a shift in how quantum algorithms are evaluated. Instead of just racing to find a good answer as quickly as possible, the focus here is on providing a verified, certified answer. The researchers recorded the entire search process, measuring the gap between the best solution found and the theoretical lower bound at every moment. This "primal-dual integral" serves as a scorecard for the algorithm's efficiency, showing exactly how quickly it narrowed down the possibilities. The study confirms that quantum sampling can be integrated into exact algorithms to provide proofs of optimality for problems that were previously only solvable by heuristics or classical methods that could not be verified by quantum means. While the range of solved problems is currently limited to length 30, which is far below the 66 achieved by the best classical methods, the proof of concept is clear: a quantum algorithm can now deliver a certificate of truth, not just a guess.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →