Given a set of agents and an equally large set of goods where each agent has some non‑negative utility for each good, is there a way to find a (fractional) perfect matching between agents and goods which is fair (envy-free) and efficient (Pareto-optimal)? This question was posed by Hylland and Zeckhauser (JPE 1979) who proved the existence of such matchings in a non-constructive way, leaving computer scientists scrambling for decades to find a polynomial time implementation of this so-called HZ mechanism until Chen et al. (SODA 2022) showed it to be PPAD-hard.
In this talk, I will show that the problem of finding any envy-free and Pareto-optimal matching is by itself PPAD-hard via an interesting construction relying on key insights about such matchings. On the other hand, I will show that the Nash bargaining mechanism which was proposed by Hosseini and Vazirani (ITCS 2022) results in 2-approximately envy-free and Pareto-optimal matchings.
This talk is based on joint work with Vijay Vazirani (EC 2024)
The Oberseminar takes place in the Seminarraum, 1st floor. Participants are invited to have coffee or tea in the lounge before.
S. Held, S. Hougardy, B. Korte, V. Traub, L. Végh, J. Vygen