Game Theory Lecture 7 Refinements: Perfect equilibrium

Similar documents
Game Theory Correlated equilibrium 1

Game Theory Lecture 10+11: Knowledge

Weak Dominance and Never Best Responses

Other Equilibrium Notions

Entropic Selection of Nash Equilibrium

Variants of Nash Equilibrium. CMPT 882 Computational Game Theory Simon Fraser University Spring 2010 Instructor: Oliver Schulte

Game Theory DR. ÖZGÜR GÜRERK UNIVERSITY OF ERFURT WINTER TERM 2012/13. Strict and nonstrict equilibria

First Prev Next Last Go Back Full Screen Close Quit. Game Theory. Giorgio Fagiolo

Computing Minmax; Dominance

Perfect Bayesian Equilibrium

Game Theory for Linguists

Refinements - change set of equilibria to find "better" set of equilibria by eliminating some that are less plausible

The WhatPower Function à An Introduction to Logarithms

Economics 201A Economic Theory (Fall 2009) Extensive Games with Perfect and Imperfect Information

Computing Minmax; Dominance

Zero-Sum Games Public Strategies Minimax Theorem and Nash Equilibria Appendix. Zero-Sum Games. Algorithmic Game Theory.

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

EC3224 Autumn Lecture #04 Mixed-Strategy Equilibrium

Unique Nash Implementation for a Class of Bargaining Solutions

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

Equilibrium Refinements

Industrial Organization, Fall 2011: Midterm Exam Solutions and Comments Date: Wednesday October

BELIEFS & EVOLUTIONARY GAME THEORY

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

SF2972 Game Theory Exam with Solutions March 15, 2013

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

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

Microeconomics for Business Practice Session 3 - Solutions

Lecture Notes on Bargaining

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

Bargaining, Contracts, and Theories of the Firm. Dr. Margaret Meyer Nuffield College

Computing Solution Concepts of Normal-Form Games. Song Chong EE, KAIST

Economics 3012 Strategic Behavior Andy McLennan October 20, 2006

Sequential Equilibria of Multi-Stage Games with Infinite Sets of Types and Actions

Belief-based Learning

Economics 703 Advanced Microeconomics. Professor Peter Cramton Fall 2017

SF2972 Game Theory Written Exam with Solutions June 10, 2011

6.254 : Game Theory with Engineering Applications Lecture 8: Supermodular and Potential Games

CHARACTERIZATIONS OF STABILITY

NTU IO (I) : Classnote 03 Meng-Yu Liang March, 2009

Iterated Strict Dominance in Pure Strategies

6 The Principle of Optimality

Supermodular Games. Ichiro Obara. February 6, 2012 UCLA. Obara (UCLA) Supermodular Games February 6, / 21

Lecture December 2009 Fall 2009 Scribe: R. Ring In this lecture we will talk about

Lecture Notes on Game Theory

Basic Game Theory. Kate Larson. January 7, University of Waterloo. Kate Larson. What is Game Theory? Normal Form Games. Computing Equilibria

Industrial Organization Lecture 3: Game Theory

Introduction to Game Theory

Approval Voting: Three Examples

SF2972 Game Theory Problem set on extensive form games

WHEN ORDER MATTERS FOR ITERATED STRICT DOMINANCE *

Advanced Microeconomics

Mathematical Economics - PhD in Economics

WEAKLY DOMINATED STRATEGIES: A MYSTERY CRACKED

Extensive Form Games I

Algorithms for cautious reasoning in games

Theory of Auctions. Carlos Hurtado. Jun 23th, Department of Economics University of Illinois at Urbana-Champaign

Game Theory: Lecture 3

The Existence of Perfect Equilibrium in Discontinuous Games

Extensive games (with perfect information)

COMPUTATION OF EXTENSIVE FORM EQUILIBRIA IN SEQUENCE FORM GAMES

Master 2 Macro I. Lecture notes #12 : Solving Dynamic Rational Expectations Models

Static (or Simultaneous- Move) Games of Complete Information

Long-Run versus Short-Run Player

Game Theory and Algorithms Lecture 2: Nash Equilibria and Examples

An axiomatization of minimal curb sets. 1. Introduction. Mark Voorneveld,,1, Willemien Kets, and Henk Norde

Game Theory: Spring 2017

Game Theory. Professor Peter Cramton Economics 300

The refined best-response correspondence in normal form games. Dieter Balkenborg, Josef Hofbauer & Christoph Kuzmics

Introduction to Game Theory Lecture Note 2: Strategic-Form Games and Nash Equilibrium (2)

1 Lattices and Tarski s Theorem

Game Theory and Rationality

Conditional equilibria of multi-stage games with infinite sets of signals and actions

DETERMINISTIC AND STOCHASTIC SELECTION DYNAMICS

ON FORWARD INDUCTION

Game Theory. Solutions to Problem Set 4

Economics 209A Theory and Application of Non-Cooperative Games (Fall 2013) Extensive games with perfect information OR6and7,FT3,4and11

A Generic Bound on Cycles in Two-Player Games

Algebra I. Course Outline

"Arrow s Theorem and the Gibbard-Satterthwaite Theorem: A Unified Approach", by Phillip Reny. Economic Letters (70) (2001),

CS 573: Algorithmic Game Theory Lecture date: Feb 6, 2008

Game Theory, Evolutionary Dynamics, and Multi-Agent Learning. Prof. Nicola Gatti

Economics 209B Behavioral / Experimental Game Theory (Spring 2008) Lecture 3: Equilibrium refinements and selection

1 The General Definition

Maskin Meets Abreu and Matsushima

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

Core-selecting package auctions. Han Dong, Hajir Roozbehani

UC Berkeley Haas School of Business Game Theory (EMBA 296 & EWMBA 211) Summer 2016

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

6.207/14.15: Networks Lecture 11: Introduction to Game Theory 3

4: Dynamic games. Concordia February 6, 2017

Game-Theoretic Learning:

Game theory lecture 4. September 24, 2012

C31: Game Theory, Lecture 1

Online Appendices for Large Matching Markets: Risk, Unraveling, and Conflation

Kalai-Smorodinsky Bargaining Solution Equilibria

Mechanism Design. Christoph Schottmüller / 27

Msc Micro I exam. Lecturer: Todd Kaplan.

Multiple Equilibria in the Citizen-Candidate Model of Representative Democracy.

Confronting Theory with Experimental Data and vice versa. European University Institute. May-Jun Lectures 7-8: Equilibrium

Transcription:

Game Theory Lecture 7 Refinements: Perfect equilibrium Christoph Schottmüller University of Copenhagen October 23, 2014 1 / 18

Outline 1 Perfect equilibrium in strategic form games 2 / 18

Refinements I multiplicity of NE some NE feel a bit unreasonable refinement : introduce some criterium that some NE satisfy and others do not most useful in extensive form games we will return to refinement later Example L R T 1,1 0,0 B 0,0 0,0 Table: (B, R) is NE; intuitive outcome? 3 / 18

Perfect equilibrium I problem with equilibrium (B, R): other player might make a mistake say, with some small probability η a player trembles then strictly better for player 1 (2) to choose T (L). 4 / 18

Perfect equilibrium II formally: G = N, (A i ), (u i ) m i : number of actions of player i tremble : vector η R m1+ +mn ++ where ηi k is probability that action ai k A i is chosen accidentally by player i define G (η) as game G but with strategy space such that S i (η) = {α i A i : α k i η k i } i.e. actions have to be completely mixed and put minimum probability ηi k on action ai k a mixed NE exists in G (η) (proof similar to the Nash theorem) 5 / 18

Perfect equilibrium III Definition (perfect equilibrium) A Nash equilibrium in G is (trembling hand) perfect if there is a sequence (η t ) t=1 of trembles converging to zero and a sequence of Nash equilibria α(η t ) in G (η t ) for each t, such that the sequence (α(η t )) t=1 converges to the Nash equilibrium in G. 6 / 18

Perfect equilibrium IV previous example: L R T 1,1 0,0 B 0,0 0,0 unique equilibrium of any G (η) is (T, L) only (T, L) is perfect equilibrium 7 / 18

Perfect equilibrium V Theorem Every finite strategic game has a perfect equilibrium. Proof. (sketch) 8 / 18

Perfect equilibrium VI Example L R T 1,1 0,0 B 1,0 2,1 Show that (T,L) is not a perfect equilibrium and (B,R) is a perfect equilibrium 9 / 18

Perfect equilibrium VII Definition (ε-perfect equilibrium) A mixed strategy profile α ε is an ε-perfect equilibrium if it is completely mixed and U i (α ε i, a i) > U i (α ε i, a i ) for any a i, a i A i implies α ε i (a i ) ε. in short: completely mixed but only best responses are played with more than ε probability 10 / 18

Perfect equilibrium VIII Theorem Let α be a mixed strategy profile in G. The following statements are equivalent 1 α is a perfect equilibrium 2 α is a limit of ε-perfect equilibria for ε going to zero 3 α is the limit of a sequence (α(ε)) ε 0 of completely mixed strategies such that α is a best response to α(ε) for ε small enough. Proof. (1) implies (2): If α is a perfect NE, there is a sequence α(η t ) of NE in G (η t ) converging to α.... (2) implies (3): 11 / 18

Perfect equilibrium IX let ai k be played with positive probability under α... (3) implies (1): Cut off if necessary the first few elements of the sequence of α(ε) such that for the remaining sequence α is a best reply to all α(ε). ηi k(ε) = α(ε)k i if α assigns zero probability to ai k ηi k(ε) = min lα(ε) l i otherwise... 12 / 18

Perfect equilibrium X Some comments: the definition says: there is a sequence ; requiring that for any sequence of trembles converging to zero there is a sequence of equilibria leads to strict perfect equilibria. However, those do often not exist: Example L M R T 1,1 1,0 0,0 B 1,1 0,0 1,0 Table: no strict perfect equilibrium L is a strictly dominant strategy Check that (T, L) and (B, L) are both perfect equilibria but not strictly perfect 13 / 18

Perfect equilibrium XI like most refinements perfection is sensitive to the formulation of the game: Example L C R T 1,1 0,0-1,-2 M 0,0 0,0 0,-2 B -2,-1-2,0-2,-2 Table: sensitive to additional irrelevant actions same game as first example plus strictly dominated actions B and R Check that (M,C) is also perfect equilibrium in this game (of course, (T,L) remains a perfect equilibrium as well) 14 / 18

Perfect equilibrium XII three player game: perfect equilibrium eliminates more than just equilibria in weekly dominated actions L R T 1,1,1 1,0,1 D 1,1,1 0,0,1 A L R T 1,0,0 0,1,2 D 0,1,0 1,0,0 Table: a NE that is not perfect despite not being in weakly dominated strategies B (D,L,A) is a NE and none of the actions is weakly dominated not perfect: for small enough trembles of P2 and P3 T will always be better than D for P1; 15 / 18

Perfect equilibrium XIII A Nash equilibrium in completely mixed strategies is also a perfect equilibrium. Why? Example H T H 1,-1-1,1 T -1,1 1,-1 Table: Matching Pennies 16 / 18

Review Questions What is the idea behind refinements? (why might we need them?) What is the idea behind perfect equilibrium? How is perfect equilibrium defined? What problems does perfect equilibrium have? reading: lecture notes by Hans Keiding p. 1-6; MSZ p. 262-267 (or OR 12.5.1) *reading: lecture notes by Hans Keiding p. 7-17 / 18

Exercises 1 Show in the game below that (T,L) is not a perfect equilibrium and (B,R) is a perfect equilibrium. L R T 2,1 1,1 M 1,2 0,2 B 0,0 3,1 2 Show that (D,L,A) is a Nash equilibrium but not a perfect equilibrium in the following 3 player game: L R T 1,0,0 1,0,1 D 1,1,1 0,1,1 A L R T 0,1,1 0,0,0 D 1,0,0 1,1,0 3 Show that no player uses a weakly dominated strategy in any perfect equilibrium. B 18 / 18