Outline for today. Stat155 Game Theory Lecture 17: Correlated equilibria and the price of anarchy. Correlated equilibrium. A driving example.

Similar documents
Strategic Games: Social Optima and Nash Equilibria

Potential Games. Krzysztof R. Apt. CWI, Amsterdam, the Netherlands, University of Amsterdam. Potential Games p. 1/3

Outline for today. Stat155 Game Theory Lecture 16: Evolutionary game theory. Evolutionarily stable strategies. Nash equilibrium.

MS&E 246: Lecture 17 Network routing. Ramesh Johari

Game Theory and Control

OIM 413 Logistics and Transportation Lecture 7: Basic Sensitivity Analysis and the Braess Paradox

MS&E 246: Lecture 4 Mixed strategies. Ramesh Johari January 18, 2007

News. Good news. Bad news. Ugly news

Game Theory Correlated equilibrium 1

Lecture 19: Common property resources

Game Theory: Spring 2017

ALGORITHMIC GAME THEORY. Incentive and Computation

Routing Games 1. Sandip Chakraborty. Department of Computer Science and Engineering, INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR.

Game Theory: introduction and applications to computer networks

Global Games I. Mehdi Shadmehr Department of Economics University of Miami. August 25, 2011

The Paradox Severity Linear Latency General Latency Extensions Conclusion. Braess Paradox. Julian Romero. January 22, 2008.

MS&E 246: Lecture 18 Network routing. Ramesh Johari

Bayesian-Nash Equilibrium

Algorithmic Game Theory. Alexander Skopalik

Game Theory for Linguists

Game Theory Lecture 10+11: Knowledge

Traffic Games Econ / CS166b Feb 28, 2012

Game Theory. Wolfgang Frimmel. Perfect Bayesian Equilibrium

Msc Micro I exam. Lecturer: Todd Kaplan.

Game Theory: Spring 2017

Parking Slot Assignment Problem

Introduction to Game Theory

EC319 Economic Theory and Its Applications, Part II: Lecture 7

AGlimpseofAGT: Selfish Routing

CS364A: Algorithmic Game Theory Lecture #13: Potential Games; A Hierarchy of Equilibria

CSC304: Algorithmic Game Theory and Mechanism Design Fall 2016

Reducing Congestion Through Information Design

EC3224 Autumn Lecture #04 Mixed-Strategy Equilibrium

Congestion Equilibrium for Differentiated Service Classes Richard T. B. Ma

Game Theory and Algorithms Lecture 2: Nash Equilibria and Examples

Parking Space Assignment Problem: A Matching Mechanism Design Approach

Game theory Lecture 19. Dynamic games. Game theory

Lecture 6: April 25, 2006

CSC304 Lecture 5. Game Theory : Zero-Sum Games, The Minimax Theorem. CSC304 - Nisarg Shah 1

Mechanism Design: Implementation. Game Theory Course: Jackson, Leyton-Brown & Shoham

RANDOM SIMULATIONS OF BRAESS S PARADOX

Polynomial-time Computation of Exact Correlated Equilibrium in Compact Games

Game Theory Fall 2003

Game Theory. Greg Plaxton Theory in Programming Practice, Spring 2004 Department of Computer Science University of Texas at Austin

Mixed Nash Equilibria

Selfish Routing. Simon Fischer. December 17, Selfish Routing in the Wardrop Model. l(x) = x. via both edes. Then,

Routing Games : From Altruism to Egoism

Level K Thinking. Mark Dean. Columbia University - Spring 2017

A (Brief) Introduction to Game Theory

Strategic Properties of Heterogeneous Serial Cost Sharing

On the Informed Principal Model with Common Values

Dynamic Atomic Congestion Games with Seasonal Flows

Recap Social Choice Functions Fun Game Mechanism Design. Mechanism Design. Lecture 13. Mechanism Design Lecture 13, Slide 1

Government 2005: Formal Political Theory I

Correlated Equilibria: Rationality and Dynamics

Extensive Form Games I

Game Theory Lecture 2

Approximate Nash Equilibria with Near Optimal Social Welfare

Game Theory Through The Computational Lens

Static (or Simultaneous- Move) Games of Complete Information

Chapter 9. Mixed Extensions. 9.1 Mixed strategies

Algorithms, Games, and Networks January 17, Lecture 2

Bayes Correlated Equilibrium and Comparing Information Structures

1 Equilibrium Comparisons

Existence of sparsely supported correlated equilibria

Correlated Equilibrium in Games with Incomplete Information

Overview. Producer Theory. Consumer Theory. Exchange

This is designed for one 75-minute lecture using Games and Information. October 3, 2006

Chapter 10 Markov Chains and Transition Matrices

Prisoner s Dilemma. Veronica Ciocanel. February 25, 2013

x 1 1 and p 1 1 Two points if you just talk about monotonicity (u (c) > 0).

MS&E 246: Lecture 12 Static games of incomplete information. Ramesh Johari

Mixed Strategies. Krzysztof R. Apt. CWI, Amsterdam, the Netherlands, University of Amsterdam. (so not Krzystof and definitely not Krystof)

6.891 Games, Decision, and Computation February 5, Lecture 2

Part IB Optimisation

Algorithmic Game Theory and Applications. Lecture 4: 2-player zero-sum games, and the Minimax Theorem

We set up the basic model of two-sided, one-to-one matching

Discrete Optimization 2010 Lecture 12 TSP, SAT & Outlook

Econometric Analysis of Games 1

Common Belief Foundations of Global Games

Lecture 10: Mechanism Design

Copyright (C) 2013 David K. Levine This document is an open textbook; you can redistribute it and/or modify it under the terms of the Creative

Discrete Optimization 2010 Lecture 12 TSP, SAT & Outlook

Pathological Outcomes of Observational Learning

Small Sample of Related Literature

4: Dynamic games. Concordia February 6, 2017

The negation of the Braess paradox as demand increases: The wisdom of crowds in transportation networks

Introduction to Game Theory

April 29, 2010 CHAPTER 13: EVOLUTIONARY EQUILIBRIUM

The discrete-time second-best day-to-day dynamic pricing scheme

Asymptotically Truthful Equilibrium Selection in Large Congestion Games

Game Theory. Professor Peter Cramton Economics 300

Game Theory. Monika Köppl-Turyna. Winter 2017/2018. Institute for Analytical Economics Vienna University of Economics and Business

Chapter 2. Equilibrium. 2.1 Complete Information Games

New Perspectives and Challenges in Routing Games: Query models & Signaling. Chaitanya Swamy University of Waterloo

Economics 201B Economic Theory (Spring 2017) Bargaining. Topics: the axiomatic approach (OR 15) and the strategic approach (OR 7).

Calibration and Nash Equilibrium

Pre-Bayesian Games. Krzysztof R. Apt. CWI, Amsterdam, the Netherlands, University of Amsterdam. (so not Krzystof and definitely not Krystof)

Introduction to Game Theory

EC319 Economic Theory and Its Applications, Part II: Lecture 2

Transcription:

Outline for today Stat55 Game Theory Lecture 7: Correlated equilibria and the price of anarchy Peter Bartlett s Example: October 5, 06 A driving example / 7 / 7 Payoff Go (-00,-00) (,-) (-,) (-,-) Nash equilibria (Go, ), (, Go), (( 0, 99 ) ( 0, 0, 0)) 99. (because we want indifference: 00p + p = p ( p)) For a two player game with strategy sets S = {,..., m} and S = {,..., n}, a correlated strategy pair is a pair of random variables (R, C) with some joint probability distribution over pairs of actions (i, j) S S. Example In the traffic signal example, Pr(, Go) = Pr(Go, ) = /. Better solution? A traffic signal: Pr((Red, Green)) = Pr((Green, Red)) = /, and both players agree: Red means, Green means Go. After they both see the traffic signal, the players have no incentive to deviate from the agreed actions. c.f. a pair of mixed strategies If we have x Sm and y Sn, then choosing the two actions (R, C) independently gives Pr(R = i, C = j) = x i y j. In the traffic signal example, we cannot have Pr(, Go) > 0 and Pr(Go, ) > 0 without Pr(Go, Go) > 0. 3 / 7 4 / 7

A correlated strategy pair for a two-player game with payoff matrices A and B is a correlated equilibrium if for all i, i S, Pr(R = i) > 0 E [a i,c R = i] E [ a i,c R = i ]. for all j, j S, Pr(C = j) > 0 E [b R,j C = j] E [ b R,j C = j ]. c.f. a Nash equilibrium A strategy profile (x, y) Sm Sn is a Nash equilibrium iff for all i, i S, Pr(R = i) > 0 E [a i,c ] E [ a i,c ]. for all j, j S, Pr(C = j) > 0 E [b R,j ] E [ b R,j ]. When R and C are independent, these expectations and the conditional expectations are identical. Thus, a Nash equilibrium is a correlated equilibrium. A correlated strategy pair for a two-player game with payoff matrices A and B is a correlated equilibrium if for all i, i S, Pr(R = i) > 0 E [a i,c R = i] E [ a i,c R = i ]. for all j, j S, Pr(C = j) > 0 E [b R,j C = j] E [ b R,j C = j ]. Traffic signal example Pr(C = Go R = ) =. Pr(C = R = ) = 0. E [a,c R = ] =, E [a Go,C R = ] = 00. 5 / 7 6 / 7 Payoff Go (-00,-00) (,-) (-,) (-,-) A correlated equilibrium Value: (0,0). 0 Nash equilibria (Go, ), value (,-). (, Go), value (-,). (( ) (, 0, 99 0 value (-,-). 0, 0)) 99, A correlated strategy pair 3 3 3 Value: (-/3,-/3). Pr(C = Go R = ) =. A correlated strategy pair E [a,c R = ] =, Pr(C = Go R = Go) = 0. 3 3 3 E [a Go,C R = ] = 99/. E [a Go,C R = Go] =, E [a,c R = Go] =. 7 / 7 8 / 7

Correlated equilibria Correlated equilibria Nash s Theorem implies there is always a correlated equilibrium. They are easy to find, via linear programming. It is not unusual for correlated equilibria to achieve better solutions for both players than Nash equilibria. Combinations of correlated equilibria Implementation We can think of a correlated equilibrium being implemented in two equivalent ways: There is a random draw of a correlated strategy pair with a known distribution, and the players see their strategy only. There is a draw of a random variable ( external event ) with a known probability distribution, and a private signal is communicated to the players about the value of the random variable. Each player chooses a mixed strategy that depends on this private signal (and the dependence is common knowledge). Given any two correlated equilibria, you can combine them to obtain another: Imagine a public random variable that determines which of the correlated equilibria will be played. Knowing which correlated equilibrium is being played, the players have no incentive to deviate. The payoffs are convex combinations of the payoffs of the two CEs. 0 / 7 9 / 7 Outline 009: New York City closed Broadway at Times Square.... with the aim of reducing traffic congestion. It was successful. Before After s Example: / 7 (NYC DOT) / 7

Example: before Other examples 005: Cheonggyecheon highway removed, speeding up traffic in downtown Seoul, South Korea. 990: 4nd Street in NYC closed for Earth Day. Traffic improved. 969: Congestion decreased in Stuttgart, West Germany, after closing a major road. Example: after Explanation? Drivers, acting rationally, seek the fastest route, which can lead to bigger delays (on average, and even for everyone). 3 / 7 4 / 7 (Karlin and Peres, 06) Mechanical and electrical versions (Nature, 99) 5 / 7 (Nature, 99) 6 / 7

Outline s Example: 7 / 7