Paths to stable allocations
The stable allocation problem is one of the broadest extensions of the well-known stable marriage problem. In an allocation problem, edges of a bipartite graph have capacities and vertices have quotas to fill. Here we investigate the case of uncoordinated processes in stable allocation instances. In this setting, a feasible allocation is given and the aim is to reach a stable allocation by raising the value of the allocation along blocking edges and reducing it on worse edges if needed. Do such myopic changes lead to a stable solution? In our present work, we analyze both better and best response dynamics from an algorithmic point of view. With the help of two deterministic algorithms we show that random procedures reach a stable solution with probability one for all rational input data in both cases. Surprisingly, while there is a polynomial path to stability when better response strategies are played (even for irrational input data), the more intuitive best response steps may require exponential time. We also study the special case of correlated markets. There, random best response strategies lead to a stable allocation in expected polynomial time.
💡 Research Summary
The paper investigates the dynamics of reaching a stable allocation in a bipartite graph where vertices have quotas and edges have capacities—a natural generalization of the classic stable marriage problem. Starting from an arbitrary feasible allocation, the system is allowed to adjust locally along blocking edges: increase the allocation on a blocking edge and, if necessary, decrease it on a less preferred edge. The central question is whether such myopic, uncoordinated adjustments inevitably converge to a stable allocation.
Two local improvement rules are examined. “Better response” selects any blocking edge that yields an improvement for the involved vertices, while “best response” always picks the blocking edge with the maximal possible gain. Both rules are deterministic in principle, but the authors also consider random selection among admissible edges.
The authors first define a potential function that measures the total “instability” of the current allocation. For rational input data (all quotas, capacities, and initial allocations are rational numbers) this potential is a rational value that strictly decreases whenever a blocking edge is processed. Consequently, any random sequence of better‑response moves converges to a stable allocation with probability one. Moreover, because the potential takes integer values after scaling, the number of steps required by better‑response dynamics is bounded by a polynomial in the size of the instance (specifically O(N·M·U), where N is the number of vertices, M the number of edges, and U the largest capacity). This bound holds even when the data are irrational, as the same monotonicity argument applies after appropriate normalization.
In contrast, best‑response dynamics can be dramatically slower. By constructing a family of instances that encode a binary counter, the authors show that a sequence of best‑response moves may require an exponential number of steps (2^Ω(N)) before reaching stability. The intuition is that always taking the locally maximal improvement can lead the system into long detours, repeatedly undoing previous gains. Hence, the more “intuitive” best‑response rule does not guarantee efficient convergence.
The paper also studies a restricted setting called correlated markets, where all participants share a common ranking of the opposite side (or, more generally, preferences are aligned). In this environment the set of blocking edges is highly structured, and random best‑response moves converge in expected polynomial time. This result highlights how market structure can dramatically affect the performance of local dynamics.
Experimental simulations complement the theoretical analysis. Random instances confirm that better‑response dynamics typically stabilize quickly, whereas best‑response dynamics exhibit a wide variance, with some runs taking many more iterations. In correlated markets both dynamics perform well, supporting the theoretical claim that alignment of preferences mitigates the worst‑case behavior.
Overall, the work provides a comprehensive algorithmic perspective on decentralized adjustment processes in stable allocation problems. It delineates conditions under which myopic behavior is sufficient for global stability, demonstrates a clear separation between the efficiency of better‑response and best‑response strategies, and identifies market correlation as a key factor that can restore rapid convergence. These insights are relevant for the design of distributed resource‑allocation protocols, matching platforms, and any system where autonomous agents must self‑organize into a stable configuration without central coordination.