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

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

Mixed Strategies; Maxmin

RECITATION 8 INTRODUCTION

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

FIRST PART: (Nash) Equilibria

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

Game Theory and Randomized Algorithms

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

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

ECON 282 Final Practice Problems

Multi-player, non-zero-sum games

(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

Mixed strategy Nash equilibrium

Computing Nash Equilibrium; Maxmin

Exercises for Introduction to Game Theory SOLUTIONS

Normal Form Games: A Brief Introduction

NORMAL FORM (SIMULTANEOUS MOVE) GAMES

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

Advanced Microeconomics (Economics 104) Spring 2011 Strategic games I

Microeconomics of Banking: Lecture 4

CMU-Q Lecture 20:

Lecture 3: Nash Equilibrium

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

Analyzing Games: Mixed Strategies

DECISION MAKING GAME THEORY

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

EC3224 Autumn Lecture #02 Nash Equilibrium

THEORY: NASH EQUILIBRIUM

PARALLEL NASH EQUILIBRIA IN BIMATRIX GAMES ISAAC ELBAZ CSE633 FALL 2012 INSTRUCTOR: DR. RUSS MILLER

Student Name. Student ID

CS510 \ Lecture Ariel Stolerman

Lecture 6: Basics of Game Theory

Economics 201A - Section 5

Lecture #3: Networks. Kyumars Sheykh Esmaili

Chapter 30: Game Theory

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.

ECO 463. SimultaneousGames

ECON 2100 Principles of Microeconomics (Summer 2016) Game Theory and Oligopoly

Game theory. Logic and Decision Making Unit 2

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

Dominant and Dominated Strategies

CSC304: Algorithmic Game Theory and Mechanism Design Fall 2016

Noncooperative Games COMP4418 Knowledge Representation and Reasoning

Chapter 3 Learning in Two-Player Matrix Games

Dominant and Dominated Strategies

Introduction to Game Theory

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.

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

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

ESSENTIALS OF GAME THEORY

Finance Solutions to Problem Set #8: Introduction to Game Theory

Introduction to IO. Introduction to IO

Homework 5 Answers PS 30 November 2013

LECTURE 26: GAME THEORY 1

Introduction to Game Theory

Problem 1 (15 points: Graded by Shahin) Recall the network structure of our in-class trading experiment shown in Figure 1

Lecture 10: September 2

Topics in Applied Mathematics

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

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

CPS 570: Artificial Intelligence Game Theory

Minmax and Dominance

Dominance and Best Response. player 2

Arpita Biswas. Speaker. PhD Student (Google Fellow) Game Theory Lab, Dept. of CSA, Indian Institute of Science, Bangalore

Game Theory Lecturer: Ji Liu Thanks for Jerry Zhu's slides

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

Lecture 7: Dominance Concepts

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

Lecture 23. Offense vs. Defense & Dynamic Games

Introduction to Game Theory

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

n-person Games in Normal Form

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

Terry College of Business - ECON 7950

Math 464: Linear Optimization and Game

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

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

Basic Solution Concepts and Computational Issues

Game Theory Week 1. Game Theory Course: Jackson, Leyton-Brown & Shoham. Game Theory Course: Jackson, Leyton-Brown & Shoham Game Theory Week 1

ECO 5341 Strategic Behavior Lecture Notes 3

Introduction to Game Theory

CSC304 Lecture 2. Game Theory (Basic Concepts) CSC304 - Nisarg Shah 1

Static or simultaneous games. 1. Normal Form and the elements of the game

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

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

EconS Game Theory - Part 1

1. Introduction to Game Theory

Computational Methods for Non-Cooperative Game Theory

Introduction to Game Theory I

Grade 7/8 Math Circles. February 14 th /15 th. Game Theory. If they both confess, they will both serve 5 hours of detention.

Dominance Solvable Games

Computing optimal strategy for finite two-player games. Simon Taylor

Game Theory. Lecture Notes By Y. Narahari. Department of Computer Science and Automation Indian Institute of Science Bangalore, India August 2012

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

Simultaneous Move Games

The extensive form representation of a game

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

Games of Perfect Information and Backward Induction

February 11, 2015 :1 +0 (1 ) = :2 + 1 (1 ) =3 1. is preferred to R iff

Game theory Computational Models of Cognition

Transcription:

Problem 1 UPenn NETS 412: Algorithmic Game Theory Game Theory Practice Bonnie Clyde Silent Confess Silent 1, 1 10, 0 Confess 0, 10 5, 5 This game is called Prisoner s Dilemma. Bonnie and Clyde have been arrested on suspicion of bank robbery, but the police do not have enough evidence to convict them without a confession. Bonnie and Clyde are separately brought to different interview rooms and asked whether or not they want to sign a confession. The police have offered them a deal, where if one confesses and the other does not, the one confessing can walk away with no time in prison and the other will serve a sentence of 10 years. If they both confess, they can split the sentence and each serve 5 years. However, Bonnie and Clyde know that the police don t have enough evidence without a confession, so if they both stay silent, they ll each only serve one year in prison on minor charges. Part a If Bonnie and Clyde weren t being interviewed separately, and could work together, which outcome would they pick? They would both stay silent, as this is the welfare-maximizing outcome. Part b Suppose that Clyde has already been interviewed and has stayed silent, and Bonnie knows this. What should Bonnie do to maximize her own utility? Why? -1. She should confess. Confessing gives her 0 utility while staying silent gives her Part c Now suppose that Clyde has already been interviewed and has confessed, and Bonnie knows this. What should Bonnie do to maximize her own utility? Why? She should confess. The payoff to staying silent is -10 compared to -5 for confessing. Part d Does Bonnie have a dominant strategy here? 1

strategy. Yes. Confess is a best-response to both of Clyde s actions, so it is a dominant Part e Does Clyde have a dominant strategy here? Yes, by a symmetric argument. Part f What is the Nash equilibrium of this game? Part g Both players confess. Why aren t there any other Nash equilibria? Both players are playing a dominant strategy it is a dominant strategy equilibrium. No other strategy can be a best response, so no other strategy can be an equilibrium. Problem 2 Alice Bob Heads Tails Heads 1, 1 1, 1 Tails 1, 1 1, 1 This game is called Matching Pennies (sometimes Odds-and-Evens ). Alice and Bob each have a penny, and they can choose to either show heads or tails. If they both show the same thing (the pennies match ), Alice wins. If they show different things, Bob wins. Part a Part b Part c Suppose Alice knows that Bob is going to show heads. What should she do? Show heads. She wins if they match. Suppose Alice knows that Bob is going to show tails. What should she do? Show tails. Again, she wins if they match. Suppose Bob knows that Alice is going to show heads. What should he do? Show tails. He wins if they don t match. 2

Part d Part e Suppose Bob knows that Alice is going to show tails. What should he do? Show heads. Again, she wins if they don t match. Does either player have a strictly dominant strategy? No. For both players, both actions are sometimes the best response to some action of the other player. Part f Is there a pure strategy Nash equilibrium? No. There is no pair of actions where both players are best-responding to each other. In other words, in any one of the outcomes, one of the players would have rather played the other choice. Part g Suppose instead of playing a strategy deterministically, each player randomizes. Suppose that Alice plays heads with probability p and Bob plays heads with probability q. Suppose that Alice knows that Bob has set q =.75. What should she set p to to maximize her utility. In other words, what is Alice s best-response to Bob s choice of q =.75? Alice should always play heads. If she does, she expects to win with probability 3/4. If she picks something smaller, say p =.9, she would win with probability.9.75+.1.25, which is less than.75. Part h Suppose Alice knows that Bob has picked q =.5. What should she set p to to maximize her utility. In other words, what is Alice s best-response to Bob s choice of q =.5? Every choice of p [0, 1] is a best-response to q =.5. This is because regardless of her choice, her expected utility is zero. If she sets p = p, she expects to win with probability p.5 + (1 p ).5 =.5. Part i If Bob has set q =.5, why does Alice picking p =.9 not result in the game being in a Nash equilibrium? Even though.9 is a best-response to.5, Bob s choice of.5 is not his best-response to Alice picking.9 (he would rather pick q = 0). A Nash equilibrium requires that every player is best-responding to every other player. Part j Why is Alice picking p =.5 and Bob picking q =.5 a Nash equilibrium? 3

We saw earlier for Alice that any choice of p is a best-response to q =.5 and by symmetry, any choice of q for Bob is a best-response to Alice picking p =.5. Since both players are best-responding to each other, this is a Nash equilibrium. Part k How do we know that this Nash equilibrium is unique? For no other pair of values of p and q are Alice and Bob best-responding to each other. If, for example, p is a little larger than.5, Bob would rather always play tails (i.e. set q = 0), but then Alice s best-response would be to always play tails (set p = 0), and now we ve moved back into pure strategy territory, and we argued earlier that there is no pure strategy equilibrium in this game. Symmetric arguments for perturbations of p and q away from.5 in other directions complete the argument. Problem 3 Both games we ve seen have had a unique equilibrium, but equilibria do not have to be unique. Here we ll analyze a game with both pure and mixed strategy equilibria. Alice Bob Party Movie Party 3, 2 1, 1 Movie 0, 0 2, 3 This game is traditionally referred to as Battle of the Sexes, but we ll call it Coordinating Activities. Alice and Bob previously discussed going to either their friend s party or to see a movie, but neither can remember what they agreed on. Alice would rather go to the party than to the movie, and Bob would rather see the movie than go to the party, but they d both rather do the same activity than end up at separate places. Part a Explain why both players going to the movie or both players going to the party are both pure strategy equilibria. At (movie,movie) and (party, party), both players are best-responding to each other. Equivalently, at these profiles, neither player wants to deviate. Part b Explain why both players randomizing over the two choices, each choosing movie with probability.5 is not a mixed strategy Nash equilibrium. If Alice is going to the movie with probability.5, then Bob s expected payoff for going to the movie is 1.5 and his expected payoff for going to the party is 1, so Bob s best response in this case is to always go to the movie, rather than mixing with probability.5. 4

Part c Explain why each player randomizing by picking their preferred activity with probability.75 is a mixed strategy Nash equilibrium. Suppose Alice picks party with probability.75, then Bob s expected payoff for going to the party is.75(2) + 0.25(0) = 1.5 and his expected payoff for going to the movie is 0.75(1) + 0.25(3) = 1.5. Since these are equal, Bob is indifferent between the choices, so any choice of a mix, including going to the movie with probability.75, is a best-response. A symmetric calculation shows that Alice is indifferent when Bob goes to the movie with probability.75, so when both players pick their preferred option with probability.75, they are indeed best-responding to each other, hence this is a Nash equilibrium. Part d Are there any other mixed strategy Nash equilibria? No, for the same reason as in Matching Pennies. For any other choice of probability for Alice, Bob will be better off playing one of his actions deterministically, and symmetrically for Alice. This has to be the only mixed strategy Nash equilibrium. Problem 4 Here s a game from class. Violin Harmonica Guitar (3,-3) (-1,1) Drums (-2,2) (1,-1) Part a What is the column player s best response to Guitar? to Drums? If the row player plays Guitar, the column player gets 3 from Violin and 1 from Harmonica, so Harmonica. Similarly, if the row player plays Drums, the column player s best response is Violin. Part b What is the column player s best response if the row player is playing p = (0.5, 0.5), that is, probability 0.5 on Guitar and 0.5 on Drums? If the column player plays Violin, she gets expected utility 0.5( 3)+0.5(2) = 0.5. If she plays Harmonica, she gets 0.5(1) + 0.5( 1) = 0. So Harmonica is the best response (putting probability 1 on Harmonica). Part c This game only has one mixed Nash equilibrium. Find it! Hint: Remember that if a player is randomizing in equilibrium, then she is indifferent between her actions, i.e. they have the same payoff. 5

See Lecture 06. 6