Popularity at Minimum Cost

We consider an extension of the {\em popular matching} problem in this paper. The input to the popular matching problem is a bipartite graph G = (A U B,E), where A is a set of people, B is a set of items, and each person a belonging to A ranks a subset of items in an order of preference, with ties allowed. The popular matching problem seeks to compute a matching M* between people and items such that there is no matching M where more people are happier with M than with M*. Such a matching M* is called a popular matching. However, there are simple instances where no popular matching exists. Here we consider the following natural extension to the above problem: associated with each item b belonging to B is a non-negative price cost(b), that is, for any item b, new copies of b can be added to the input graph by paying an amount of cost(b) per copy. When G does not admit a popular matching, the problem is to “augment” G at minimum cost such that the new graph admits a popular matching. We show that this problem is NP-hard; in fact, it is NP-hard to approximate it within a factor of sqrt{n1}/2, where n1 is the number of people. This problem has a simple polynomial time algorithm when each person has a preference list of length at most 2. However, if we consider the problem of “constructing” a graph at minimum cost that admits a popular matching that matches all people, then even with preference lists of length 2, the problem becomes NP-hard. On the other hand, when the number of copies of each item is “fixed”, we show that the problem of computing a minimum cost popular matching or deciding that no popular matching exists can be solved in O(mn1) time, where m is the number of edges.

💡 Research Summary

The paper introduces a cost‑aware extension of the classic popular matching problem. In the standard setting we are given a bipartite graph (G=(A\cup B,E)) where (A) denotes people and (B) denotes items. Each person (a\in A) ranks a subset of items, ties being allowed. A matching (M) is called popular if there is no other matching (M’) that is preferred by a larger number of people. While many instances admit a popular matching, there are simple graphs for which none exists.

To remedy this, the authors associate a non‑negative price (\text{cost}(b)) with every item (b\in B). By paying (\text{cost}(b)) one may create an additional copy of (b) (and consequently new incident edges). The central computational question becomes: Given a graph that does not admit a popular matching, add copies of items at minimum total cost so that the resulting graph does admit a popular matching.

The paper studies two closely related variants:

1. Augmentation problem – we only require that the augmented graph possesses some popular matching.
2. Full‑matching problem – we require a popular matching that matches all people (i.e., a perfect popular matching).

Main Theoretical Results

• NP‑hardness of augmentation.
  By a reduction from the classic Vertex Cover problem the authors prove that deciding whether a given budget suffices to obtain a popular matching is NP‑complete. Moreover, they show a strong inapproximability: unless P=NP, no polynomial‑time algorithm can achieve an approximation factor better than (\sqrt{n_1}/2), where (n_1=|A|) is the number of people. This hardness holds even when costs are uniform (every copy costs 1).

• Polynomial‑time algorithm for bounded preference lists.
  When each person’s preference list contains at most two items, the augmentation problem becomes tractable. The authors model the decision as a minimum‑cost vertex cover on a derived auxiliary graph and solve it via a min‑cost flow construction. The algorithm runs in (O(m n_1)) time, where (m) is the total number of declared preferences (edges).

• NP‑hardness of the full‑matching variant even with lists of length two.
  Adding the requirement that every person must be matched dramatically raises the difficulty. Using a reduction from Exact‑3‑Set‑Cover, the paper shows that finding the cheapest set of copies that yields a perfect popular matching remains NP‑complete, even when each person likes at most two items.

• Fixed‑copy case.
  If the number of copies allowed for each item is part of the input (i.e., a hard bound (k_b) for each (b)), the problem simplifies. One can first instantiate the prescribed copies, then run any standard popular‑matching algorithm. The authors present an (O(m n_1)) procedure that either finds a minimum‑cost popular matching under the fixed‑copy constraints or correctly reports that none exists.

Technical Approach

The hardness proofs rely on encoding combinatorial covering problems into the popularity condition. In the Vertex‑Cover reduction, each vertex of the original graph becomes an item, each edge becomes a person who prefers the two incident items, and a copy of an item corresponds to selecting that vertex in the cover. A popular matching exists after augmentation exactly when the selected copies form a vertex cover. The approximation lower bound follows from known hardness of approximating Vertex Cover.

For the two‑preference case, the authors observe that the underlying graph is a collection of disjoint paths and cycles. Popularity can be characterized by a simple parity condition on each component. By constructing a flow network where source edges represent people, sink edges represent items, and intermediate edges carry the copy‑selection cost, the minimum‑cost flow yields the optimal set of items to duplicate.

The Exact‑3‑Set‑Cover reduction for the full‑matching variant introduces “forced” people that must be matched to newly created copies, thereby enforcing that each selected copy corresponds to a set in the cover. The reduction preserves the two‑preference restriction, establishing NP‑hardness under very tight constraints.

Implications and Applications

The model captures realistic scenarios where items (e.g., seats in a class, hospital beds, advertisement slots) can be provisioned at a known marginal cost, and a platform wishes to guarantee a socially acceptable allocation (popular matching) while spending as little as possible. The dichotomy between the tractable bounded‑preference case and the intractable general case suggests that system designers should either limit the length of preference lists or accept heuristic solutions when the problem size grows.

Furthermore, the fixed‑copy result provides a practical algorithm for settings where inventory limits are known in advance (e.g., a limited number of identical products). The (O(m n_1)) runtime is comparable to classic matching algorithms, making it feasible for large‑scale platforms.

Conclusion

The paper establishes a new frontier at the intersection of matching theory and cost optimization. It proves that augmenting a graph to admit a popular matching is computationally hard in general, yet reveals a sweet spot—preference lists of length two—where exact polynomial‑time solutions exist. The work also highlights that demanding a perfect popular matching re‑introduces hardness even under the same restriction. By providing both hardness results and efficient algorithms for important special cases, the authors lay a solid theoretical foundation for future research on cost‑aware popular matchings and their deployment in real‑world allocation systems.