Главная
Study mode:
on
1
Intro
2
Rank-k games
3
Suggested example of a rank-1 game
4
NE of games with parameter X
5
Global Newton Method [Govindan/Wilson 2003]
6
Rank reduction by intersection with hyperplane
7
works for rank 1: monotone path, unique rank-O NE
8
LP for zero-sum game
9
LP for parameterized zero-sum game
10
Parameterized matrix (column bonuses)
11
Equivalent: Parameterized objective function
12
"Wrap-around" hyperplane defines the path
13
Algorithm 1: Binary search
14
Algorithm 2: Enumeration
15
Parameterized LP [Murty 1980]
16
Inspired by Murty
17
Output efficiency
18
PPAD- versus NP-hardness
Description:
Explore algorithms for rank-1 bimatrix games in this lecture from the Hausdorff Trimester Program on Combinatorial Optimization. Delve into the concept of game rank, defined as the matrix rank of the sum of two payoff matrices, and discover how rank-1 games can be solved efficiently. Learn about polynomial-time algorithms for finding Nash equilibria and the application of path-following techniques. Examine the potential for exponential numbers of equilibria in rank-1 games and consider the conjecture regarding the co-NP-completeness of uniqueness in Nash equilibrium. Investigate topics such as rank reduction, global Newton method, parameterized zero-sum games, and the "wrap-around" hyperplane concept. Gain insights into binary search and enumeration algorithms, as well as parameterized linear programming inspired by Murty's work. Conclude by discussing the implications of PPAD- versus NP-hardness in the context of rank-1 bimatrix games.

Algorithms for Rank-1 Bimatrix Games

Hausdorff Center for Mathematics
Add to list
0:00 / 0:00