New Algorithm Balances Utility in Stable Matchings with a Twist

Stable matching problems, famously solved by Gale and Shapley, have long focused on stability alone. Now a new paper from Yao Sheng and Yu Yokoi tackles fairness: the Stable Matching Problem with Minimum Utility Gap. Their algorithm aims to produce a stable matching where the most and least satisfied agents are as close as possible in utility, measured either by absolute difference or ratio. The key insight is that the set of all stable matchings can be represented as a lattice using "rotations" — and each agent's utility is affected by a chain of these rotations. This structure enables a brute-force search over a polynomial number of candidate matchings, yielding an efficient solution for many-to-many settings with arbitrary preferences.

The work also clarifies how this objective relates to existing optimization frameworks. The authors show that minimizing the utility gap is not reducible to minimum-cut formulations, a common technique for stable matching with costs. However, they identify a special case (when agents can be divided into two groups with identical preferences within each group) where the problem becomes submodular function minimization — another tractable class. This positions the result as a new branch of computational social choice, with potential applications in labor markets, school assignment, and kidney exchange where fairness is paramount. The paper is available on arXiv (2607.07160) and is under review for publication.