Game Theory. 4: Nash equilibrium in different games and mixed strategies

Similar documents
Game Theory. 4: Nash equilibrium in different games and mixed strategies

Multi-player, non-zero-sum games

Game theory Computational Models of Cognition

NORMAL FORM (SIMULTANEOUS MOVE) GAMES

1\2 L m R M 2, 2 1, 1 0, 0 B 1, 0 0, 0 1, 1

Computing Nash Equilibrium; Maxmin

Microeconomics of Banking: Lecture 4

Mixed Strategies; Maxmin

Finite games: finite number of players, finite number of possible actions, finite number of moves. Canusegametreetodepicttheextensiveform.

Topics in Applied Mathematics

Distributed Optimization and Games

Econ 302: Microeconomics II - Strategic Behavior. Problem Set #5 June13, 2016

Analyzing Games: Mixed Strategies

Introduction to (Networked) Game Theory. Networked Life NETS 112 Fall 2016 Prof. Michael Kearns

Introduction to (Networked) Game Theory. Networked Life NETS 112 Fall 2014 Prof. Michael Kearns

Lecture 23. Offense vs. Defense & Dynamic Games

Lecture 6: Basics of Game Theory

Lecture 3: Nash Equilibrium

THEORY: NASH EQUILIBRIUM

CMU Lecture 22: Game Theory I. Teachers: Gianni A. Di Caro

Section Notes 6. Game Theory. Applied Math 121. Week of March 22, understand the difference between pure and mixed strategies.

UPenn NETS 412: Algorithmic Game Theory Game Theory Practice. Clyde Silent Confess Silent 1, 1 10, 0 Confess 0, 10 5, 5

Computational Methods for Non-Cooperative Game Theory

Note: A player has, at most, one strictly dominant strategy. When a player has a dominant strategy, that strategy is a compelling choice.

EC3224 Autumn Lecture #02 Nash Equilibrium

Computational Aspects of Game Theory Bertinoro Spring School Lecture 2: Examples

Advanced Microeconomics (Economics 104) Spring 2011 Strategic games I

LECTURE 26: GAME THEORY 1

U strictly dominates D for player A, and L strictly dominates R for player B. This leaves (U, L) as a Strict Dominant Strategy Equilibrium.

Prisoner 2 Confess Remain Silent Confess (-5, -5) (0, -20) Remain Silent (-20, 0) (-1, -1)

Introduction to Game Theory

Lecture #3: Networks. Kyumars Sheykh Esmaili

Distributed Optimization and Games

Game Theory ( nd term) Dr. S. Farshad Fatemi. Graduate School of Management and Economics Sharif University of Technology.

37 Game Theory. Bebe b1 b2 b3. a Abe a a A Two-Person Zero-Sum Game

Session Outline. Application of Game Theory in Economics. Prof. Trupti Mishra, School of Management, IIT Bombay

Games. Episode 6 Part III: Dynamics. Baochun Li Professor Department of Electrical and Computer Engineering University of Toronto

CMU-Q Lecture 20:

Lecture 24. Extensive-Form Dynamic Games

Game Theory and Economics Prof. Dr. Debarshi Das Humanities and Social Sciences Indian Institute of Technology, Guwahati

Mixed strategy Nash equilibrium

Introduction to IO. Introduction to IO

ECO 5341 Strategic Behavior Lecture Notes 3

Nash Equilibrium. An obvious way to play? Player 1. Player 2. Player 2

1. Simultaneous games All players move at same time. Represent with a game table. We ll stick to 2 players, generally A and B or Row and Col.

Game Theory. Wolfgang Frimmel. Subgame Perfect Nash Equilibrium

Spring 2014 Quiz: 10 points Answer Key 2/19/14 Time Limit: 53 Minutes (FAS students: Teaching Assistant. Total Point Value: 10 points.

Nash Equilibrium. Felix Munoz-Garcia School of Economic Sciences Washington State University. EconS 503

Game Theory and Randomized Algorithms

Copyright 2008, Yan Chen

ESSENTIALS OF GAME THEORY

8.F The Possibility of Mistakes: Trembling Hand Perfection

Lecture 13(ii) Announcements. Lecture on Game Theory. None. 1. The Simple Version of the Battle of the Sexes

Normal Form Games. Here is the definition of a strategy: A strategy is a complete contingent plan for a player in the game.

Terry College of Business - ECON 7950

Economics 201A - Section 5

Backward Induction and Stackelberg Competition

Basic Game Theory. Economics Auction Theory. Instructor: Songzi Du. Simon Fraser University. September 7, 2016

CS510 \ Lecture Ariel Stolerman

(a) Left Right (b) Left Right. Up Up 5-4. Row Down 0-5 Row Down 1 2. (c) B1 B2 (d) B1 B2 A1 4, 2-5, 6 A1 3, 2 0, 1

Exploitability and Game Theory Optimal Play in Poker

Game theory. Logic and Decision Making Unit 2

Math 464: Linear Optimization and Game

Summary Overview of Topics in Econ 30200b: Decision theory: strong and weak domination by randomized strategies, domination theorem, expected utility

Introduction to Game Theory

CSCI 699: Topics in Learning and Game Theory Fall 2017 Lecture 3: Intro to Game Theory. Instructor: Shaddin Dughmi

Strategic Bargaining. This is page 1 Printer: Opaq

4/21/2016. Intermediate Microeconomics W3211. Lecture 20: Game Theory 2. The Story So Far. Today. But First.. Introduction

Weeks 3-4: Intro to Game Theory

Game Theory two-person, zero-sum games

Adversarial Search and Game Theory. CS 510 Lecture 5 October 26, 2017

FIRST PART: (Nash) Equilibria

Chapter 3 Learning in Two-Player Matrix Games

Game Theory Refresher. Muriel Niederle. February 3, A set of players (here for simplicity only 2 players, all generalized to N players).

1. Introduction to Game Theory

player 1 x>y>z x>z>y y>x>z y>z>x z>y>x z>x>y not aggressive aggressive most aggressive

DECISION MAKING GAME THEORY

Homework answers IEE1149 Summer 2015

Lect 15:Game Theory: the math of competition

Introduction to Game Theory

final examination on May 31 Topics from the latter part of the course (covered in homework assignments 4-7) include:

Lecture 7: Dominance Concepts

Terry College of Business - ECON 7950

ECON 282 Final Practice Problems

CHAPTER LEARNING OUTCOMES. By the end of this section, students will be able to:

Game Theory: Introduction. Game Theory. Game Theory: Applications. Game Theory: Overview

Noncooperative Games COMP4418 Knowledge Representation and Reasoning

ECO 220 Game Theory. Objectives. Agenda. Simultaneous Move Games. Be able to structure a game in normal form Be able to identify a Nash equilibrium

Introduction Economic Models Game Theory Models Games Summary. Syllabus

Game Theory: The Basics. Theory of Games and Economics Behavior John Von Neumann and Oskar Morgenstern (1943)

Appendix A A Primer in Game Theory

Student Name. Student ID

Rationality and Common Knowledge

Game Theory -- Lecture 6. Patrick Loiseau EURECOM Fall 2016

Reading Robert Gibbons, A Primer in Game Theory, Harvester Wheatsheaf 1992.

NORMAL FORM GAMES: invariance and refinements DYNAMIC GAMES: extensive form

Lecture Notes on Game Theory (QTM)

Domination Rationalizability Correlated Equilibrium Computing CE Computational problems in domination. Game Theory Week 3. Kevin Leyton-Brown

0.1 Battle of the Sexes. 0.2 Chicken. 0.3 Coordination Game

Introduction: What is Game Theory?

Transcription:

Game Theory 4: Nash equilibrium in different games and mixed strategies

Review of lecture three A game with no dominated strategy: The battle of the sexes The concept of Nash equilibrium The formal definition of NE How to find NE in matrix games 2

Which side of the road? Mr. Green Mr. Red LEFT RIGHT LEFT (1, 1) (-1, -1) RIGHT (-1, -1) (1, 1) You have two NE: (LEFT, LEFT) and (RIGHT, RIGHT) This game is a coordination one In a coordination game the problem is not to cheat or lie but to find out the way to get the mutual benefit by coordinating their actions (i.e., choosing the same strategy) How to do that? 3

Focal points For games of coordination this can be done by finding some elements of interaction that appears in some way as prominent to players These aspects are called focal points 4

Examples (T. Schelling 1960) Name a city of UK. If you all name the same, you win a prize You are to meet somebody in Milan for a reason that you both consider very important. You were not been told where to meet and you cannot communicate. Guess where to go Ask to Peppino!!! 5

Totò, Peppino e la Malafemmina: http://www.youtube.com/watch?v=im2gpqwk8uk 6

Examples (T. Schelling 1960) people can often concert their intentions or expectations with others if each knows that the other is trying to do the same Focal points impose themselves on the players attention for reasons that the formal theory overlooks 7

Application to BS: asymmetric coordination game fight Ann ballet Bert fight (2,1) (0,0) ballet (0,0) (1,2) 1. Today is Ann birthday 2. Last evening out Ann and Bert went to a ballet Still it may happen that players do not find the way to reach one of their preferred results coordination failure 8

An application to Politics Nuclear arms race (first case) Both US and USSR have two strategies: ARM and REFRAIN Suppose both US and USSR care only for military supremacy (A vs. R)>(R vs. R)>(A vs. A)>(R vs. A) 9

Tough superpowers arms race US USSR REFRAIN ARM REFRAIN (3, 3) (1, 4) ARM (4, 1) (2, 2) By choosing ARM (leading to (2,2) both players are worse off than if they could reach an arm control agreement, leading to (3,3) However this outcome is unstable (it is not a Nash Equilibrium) The only NE is (ARM,ARM) leading to (2,2) the game is a PD 10

Nuclear arms race (second case) US and USSR have the same two strategies: ARM and REFRAIN But their leadership care also for military expenditures that reduce people s standard of living However security is worth more than expenditures (R vs. R)>(A vs.r)>(a vs. A)>(R vs. A) the game becomes a particular coordination game 11

Mild superpowers arms race US USSR REFRAIN ARM REFRAIN (4, 4) (1, 3) ARM (3, 1) (2, 2) Two NE (ARM,ARM) and (REFRAIN,REFRAIN) 12

Assurance game US USSR REFRAIN ARM REFRAIN (4, 4) (1, 3) ARM (3, 1) (2, 2) (REFRAIN,REFRAIN) is better for both but difficult to reach If one player has reason to think that the other chooses ARM, it too will choose ARM To choose REFRAIN a player needs the assurance that the other will do the same assurance game In the case of superpowers this assurance would have been a mutual control However they never accepted it: sometimes pessimism 13 breads pessimism

Nuclear arms race (third case) Superpowers acknowledge the situation has become dramatic (the Cuban crisis?) Both assume having two strategies: send an ULTIMATUM or RETRAIT Double U brings about a nuclear conflict (the worst case for both) The best result is to send U when the other plays R The second result is the double R R against U is the third result (U vs. R)>(R vs. R)>(R vs. U)>(U vs. U) 14

Ultimatum game US USSR R U R (2, 2) (1, 3) U (3, 1) (0, 0) Two NE: (U,R) (3,1) and (R,U) (1,3) This game also called a chicken game: people do not have to coordinate on the same strategy! These possible solutions ask (ONCE AGAIN!) for a richer knowledge of the external environment Which one of the two NE will be chosen depends on possible strategic moves (i.e., credible precommitments) 15

Strategic moves A player may take an initiative that influences the other player s choice in a way favorable to the first one One can constrain the opponent s choice by constraining one s own behavior Bert can arrive home with the tickets for fight so that the choice ballet is implicitly cancelled A military commander can order his guard to burn the bridge of the river just passed so that his army knows that can never retreat and must fight fiercely (Sun Tsu, The art of war) 16

Exercise Take a look at: http://www.youtube.com/watch?v=fn7d_a0pmio&feat ure=related A Chicago teenager called Ren moves to a small city in Iowa. Ren s love of dancing and partying causes friction with the community. Much of the movie centers on the competition between Ren and the local tough guy named Chuck At one stage Chuck challanges Ren to a tractor face-off. In this face-off Ren and Chuck have to drive tractors directly at each other. Whoever swerves out of the way first is considered a chicken Represent the game and solve it!

Matching pennies Two players: A and B own a coin each, turned secretly on head or tail Confronting coins, if both show the same face A takes both; otherwise B takes both A B head tail head (1, 1) ( 1, 1) tail ( 1, 1) (1, 1) A zero-sum game with no NE. What to do? 18

Mixed strategy Every finite game (having a finite number of players and a finite strategy space) has at least one NE (in pure OR in mixed strategies) A mixed strategy for a player is a probability distribution over her (pure) strategies An example: A (½, ½) is a possible mixed strategy. As well as A (2/3, 1/3), or (1/4, ¾) A mixed strategy include also all pure strategies (when the probability of a strategy is = 1 and the probability for the other strategy is = 0, i.e., A (1, 0) ) 19

Mixed strategy What is a mixed-strategy Nash Equilibrium (MSNE)? A MSNE is a profile of MS T*ϵ T s.t. u i (T i *, T _i *) ui(t i, T _i *) i and T i ϵ T How to estimate a MSNE? Let s guess that A mixed between H and T. If this strategy is optimal for A (in response to the other player s strategy), then it must be that the expected payoff from playing H equals the expected playoff from playing T. Why that? Otherwise, player A would strictly prefer to pick either H or T (i.e., playing a pure strategy) 20

Mixed strategy But how can player A s strategies H and T yield the same expected payoff? It must be that player B s behavior generates this expectation (because if B played a pure strategy, then A would strictly prefer one of its strategies over the other but then also B would prefer to change her pure strategy and so on ) 21

Mixed strategy Let us call p the probability for A to play head and 1 p her probability to play tail Let us call q the probability for B to play head and 1 q his probability to play tail So how to proceed? 22

The calculus way EU A (H q) = q-1+q = 2q-1 EU A (T q) = -q+1-q=1-2q EU A (H q) = EU A (T q) implies q=1/2 Similarly: EU B (H p) = -p+1-p = 1-2p EU B (T p) = p-1+p=2p-1 EU B (H p) = EU B (T p) implies p=1/2 The mixed strategy profile ((1/2, ½), (1/2, ½)) is a MSNE 23

The calculus way Given player B s mixed strategy (1/2, ½), player A s mixed strategy (1/2, ½) is a best response, and vice-versa In fact, every strategy is a best response for player A, given player B s mixed strategy: i.e., (3/4, 1/4) (0, 1) (1, 0) In this sense, if player A changes his behavior, given player B s mixed strategy, it does not worse his situation In a pure NE, on the contrary, if you deviate from your equilibrium strategy, you always worse your situation: so you return pretty fast to that equilibrium! As a resutl: a MSNE is a weaker solution than a pure NE, but it is still an equilibrium, i.e., the solution to a strategic interdependent situation (and in some cases, the only solution available ) 24

A graphical way Note that constructing a MSNE entails an interesting new twist: we look for a mixed strategy for one player that makes the other player indifferent between her pure strategies. This is the best method of calculating MSNE A graphical way to find a MSNE Mutual Best Responses! 25

The Battle of the Sexes reprise Man Woman Football Opera Football (3, 2) (1, 1) Opera (0, 0) (2, 3) Man and Woman like each other, but Man of course likes football more than Opera They have too coordinate their behavior There are two pure NE and one MSNE Find them! Compared to a pure NE, a MSNE is less stable 26

An interpretation of MSNE Repeated game interpretation: the probabilities identified by a MSNE correspond to the frequencies of times that each strategy is played by each player over time in equilibrium Evolutionary game interpretation: the probabilities identified by a MSNE correspond to the percentage of players playing each pure strategy in a given population in equilibrium 27

The World War I game (Homework) Consider the following scenario The British are deciding whether to attack Germany at the Somme river in France or to attack Germany s ally Turkey at Constantinople. The Somme is closer to German territory so a big victory there will end the war sooner that a breakthrough against Turkey The Germans must decide whether to concentrate their defensive forces at the Somme or bolster Turkey If the attacks comes where the defense is strong, the attack will fail. If the attack happens where the defense is weak, the attackers win 28

NASH EQUILIBRIUM IAND MIXED STRATEGIES The World War I game (Homework) Assume that British preferences are given by u B (victory at the Somme) = 2 > u B (Victory in Turkey)=1 > u B (losing either place) = 0 and that the preferences of the Germans are given by u G (successful defense) = 2 > u G (lose in Turkey) = 1 > u G (lose at the Somme) = 0 Further assume that the British strategy space is (attack the Somme, attack Turkey) and that the Germany strategy space is (defend the Somme, defend Turkey) So: a) represent this game in Matrix form; b) find all the pure strategy and mixed strategy NE of this game, using both methods discussed (i.e., including also drawing best reply correspondences) 29