Seminars
View all Seminars | Download ICal for this eventQuery-Efficient Algorithms to Find the Unique Nash Equilibrium in a Two-Player Zero-Sum Matrix Game
Series: Bangalore Theory Seminars
Speaker: Arnab Maiti, University of Washington
Date/Time: Feb 23 11:00:00
Location: Online Talk (See Teams link below)
Abstract:
We study the query complexity of identifying Nash equilibria in two-player zero-sum matrix games. Grigoriadis and Khachiyan (1995) showed that any deterministic algorithm needs to query Ω(n2) entries in worst case from an n ? n input matrix in order to compute an ε-approximate Nash equilibrium, where ε < 1/2 . Moreover, they designed a randomized algorithm that queries O(nlog(n)/ε2) entries from the input matrix in expectation and returns an ε-approximate Nash equilibrium when the entries of the matrix are bounded between ??1 and 1. However, these two results do not completely characterize the query complexity of finding an exact Nash equilibrium in two-player zero-sum matrix games. In this work, we characterize the query complexity of finding an exact Nash equilibrium for two-player zero-sum matrix games that have a unique Nash equilibrium (x*,y*). We first show that any randomized algorithm needs to query Ω(nk) entries of the input matrix A ?? Rn?n in expectation in order to find the unique Nash equilibrium where k = |supp(x*)|. We complement this lower bound by presenting a simple randomized algorithm that, with probability 1??δ, returns the unique Nash equilibrium by querying at most O(nk4 · polylog(n/δ)) entries of the input matrix A ?? Rn?n. In the special case when the unique Nash Equilibrium is a pure-strategy Nash equilibrium (PSNE), we design a simple deterministic algorithm that finds the PSNE by querying at most O(n) entries of the input matrix.
Microsoft teams link:
Link
We are grateful to the Kirani family for generously supporting the theory seminar series
Hosts: KVN Sreenivas, Rameesh Paul, Rahul Madhavan, Manisha Padala