Standard and Extended Form Games A Lesson in Multiagent System Based on Jose Vidal’s book Fundamentals of Multiagent Systems Henry Hexmoor, SIUC.

Slides:



Advertisements
Similar presentations
Chapter 17: Making Complex Decisions April 1, 2004.
Advertisements

Vincent Conitzer CPS Repeated games Vincent Conitzer
Nash’s Theorem Theorem (Nash, 1951): Every finite game (finite number of players, finite number of pure strategies) has at least one mixed-strategy Nash.
This Segment: Computational game theory Lecture 1: Game representations, solution concepts and complexity Tuomas Sandholm Computer Science Department Carnegie.
Mixed Strategies CMPT 882 Computational Game Theory Simon Fraser University Spring 2010 Instructor: Oliver Schulte.
Game Theory S-1.
Two-Player Zero-Sum Games
Non-Cooperative Game Theory To define a game, you need to know three things: –The set of players –The strategy sets of the players (i.e., the actions they.
6-1 LECTURE 6: MULTIAGENT INTERACTIONS An Introduction to MultiAgent Systems
Chapter 14 Infinite Horizon 1.Markov Games 2.Markov Solutions 3.Infinite Horizon Repeated Games 4.Trigger Strategy Solutions 5.Investing in Strategic Capital.
Operations Research Assistant Professor Dr. Sana’a Wafa Al-Sayegh 2 nd Semester ITGD4207 University of Palestine.
The basics of Game Theory Understanding strategic behaviour.
Chapter 6 Game Theory © 2006 Thomson Learning/South-Western.
Chapter 6 Game Theory © 2006 Thomson Learning/South-Western.
An Introduction to... Evolutionary Game Theory
1 Game Theory. 2 Agenda Game Theory Matrix Form of a Game Dominant Strategy and Dominated Strategy Nash Equilibrium Game Trees Subgame Perfection.
Negotiation A Lesson in Multiagent System Based on Jose Vidal’s book Fundamentals of Multiagent Systems Henry Hexmoor SIUC.
EKONOMSKA ANALIZA PRAVA. Game Theory Outline of the lecture: I. What is game theory? II. Elements of a game III. Normal (matrix) and Extensive (tree)
Game-theoretic analysis tools Necessary for building nonmanipulable automated negotiation systems.
Chapter 11 Game Theory and the Tools of Strategic Business Analysis.
Games What is ‘Game Theory’? There are several tools and techniques used by applied modelers to generate testable hypotheses Modeling techniques widely.
An Introduction to Game Theory Part I: Strategic Games
5/16/20151 Game Theory Game theory was developed by John Von Neumann and Oscar Morgenstern in Economists! One of the fundamental principles of.
Eponine Lupo.  Game Theory is a mathematical theory that deals with models of conflict and cooperation.  It is a precise and logical description of.
Slide 1 of 13 So... What’s Game Theory? Game theory refers to a branch of applied math that deals with the strategic interactions between various ‘agents’,
A camper awakens to the growl of a hungry bear and sees his friend putting on a pair of running shoes, “You can’t outrun a bear,” scoffs the camper. His.
Chapter 12 Choices Involving Strategy McGraw-Hill/Irwin Copyright © 2008 by The McGraw-Hill Companies, Inc. All Rights Reserved.
Algoritmi per Sistemi Distribuiti Strategici
Game-Theoretic Approaches to Multi-Agent Systems Bernhard Nebel.
Basics on Game Theory For Industrial Economics (According to Shy’s Plan)
Review: Game theory Dominant strategy Nash equilibrium
Advanced Microeconomics Instructors: Wojtek Dorabialski & Olga Kiuila Lectures: Mon. & Wed. 9:45 – 11:20 room 201 Office hours: Mon. & Wed. 9:15 – 9:45.
Artificial Intelligence for Games and Puzzles1 Games in the real world Many real-world situations and problems.
APEC 8205: Applied Game Theory Fall 2007
An introduction to game theory Today: The fundamentals of game theory, including Nash equilibrium.
UNIT II: The Basic Theory Zero-sum Games Nonzero-sum Games Nash Equilibrium: Properties and Problems Bargaining Games Bargaining and Negotiation Review.
QR 38, 2/22/07 Strategic form: dominant strategies I.Strategic form II.Finding Nash equilibria III.Strategic form games in IR.
Introduction to Game Theory Yale Braunstein Spring 2007.
Artificial Intelligence for Games and Puzzles1 Games in the real world Many real-world situations and.
UNIT II: The Basic Theory Zero-sum Games Nonzero-sum Games Nash Equilibrium: Properties and Problems Bargaining Games Bargaining and Negotiation Review.
Minimax strategies, Nash equilibria, correlated equilibria Vincent Conitzer
Strategic Game Theory for Managers. Explain What is the Game Theory Explain the Basic Elements of a Game Explain the Importance of Game Theory Explain.
Game representations, solution concepts and complexity Tuomas Sandholm Computer Science Department Carnegie Mellon University.
Monetary Economics Game and Monetary Policymaking.
Dynamic Games of complete information: Backward Induction and Subgame perfection - Repeated Games -
Game-theoretic analysis tools Tuomas Sandholm Professor Computer Science Department Carnegie Mellon University.
Game Theory: introduction and applications to computer networks Game Theory: introduction and applications to computer networks Lecture 2: two-person non.
Chapters 29, 30 Game Theory A good time to talk about game theory since we have actually seen some types of equilibria last time. Game theory is concerned.
Ásbjörn H Kristbjörnsson1 The complexity of Finding Nash Equilibria Ásbjörn H Kristbjörnsson Algorithms, Logic and Complexity.
Lecture 5 Introduction to Game theory. What is game theory? Game theory studies situations where players have strategic interactions; the payoff that.
1 What is Game Theory About? r Analysis of situations where conflict of interests is present r Goal is to prescribe how conflicts can be resolved 2 2 r.
Mohsen Afsharchi Multiagent Interaction. What are Multiagent Systems?
Intermediate Microeconomics Game Theory. So far we have only studied situations that were not “strategic”. The optimal behavior of any given individual.
Zero-sum Games The Essentials of a Game Extensive Game Matrix Game Dominant Strategies Prudent Strategies Solving the Zero-sum Game The Minimax Theorem.
5.1.Static Games of Incomplete Information
Lec 23 Chapter 28 Game Theory.
By: Donté Howell Game Theory in Sports. What is Game Theory? It is a tool used to analyze strategic behavior and trying to maximize his/her payoff of.
Game Theory Dr. Andrew L. H. Parkes “Economics for Business (2)” 卜安吉.
Game theory basics A Game describes situations of strategic interaction, where the payoff for one agent depends on its own actions as well as on the actions.
Chapter 28 Game Theory.
Vincent Conitzer CPS Repeated games Vincent Conitzer
Multiagent Systems Game Theory © Manfred Huber 2018.
Game Theory Chapter 12.
Presented By Aaron Roth
Learning 6.2 Game Theory.
Multiagent Systems Repeated Games © Manfred Huber 2018.
Vincent Conitzer Repeated games Vincent Conitzer
Normal Form (Matrix) Games
Vincent Conitzer CPS Repeated games Vincent Conitzer
Presentation transcript:

Standard and Extended Form Games A Lesson in Multiagent System Based on Jose Vidal’s book Fundamentals of Multiagent Systems Henry Hexmoor, SIUC

Standard and Extended Form Games In all multiagent systems we have a set of autonomous agents each performing its own actions using whatever information is has available. The other agents are also taking actions, each agent must also take these into account when deciding what to do. The agent must decide what to do when their choice of action depends on the others’ choices A set of mathematical tools has been developed over the years to model and solve these problems. It is known as game theory

Origin of Game Theory It was first formally introduced in the book “The Theory of Games and Economic Behavior”, (Neumann and Morgenstern, 1944). Some game are noncooperative because the agents’ preferred sets of actions can be in conflict with each other. That is, what is good for one agent might be bad for the others.

Standard and Extended Form Games Two most basic forms of games: normal and extended form games. These games are known to be non- cooperative games because the agents’ preferred sets of actions can be in conflict with each other.

Games in Normal Form Normal form, also known as strategic form. In the simplest type of game we have two agents each of which must take one of two possible actions. The agents take their actions at the same time. They will then each receive a utility value, or payoff, based on their joint actions. Games such as this one can be represented using a payoff matrix

Games in Normal Form Normal form games also assume that the players have common knowledge of the utilities that all players can receive. players have common knowledge of the utilities that all players can receive. That is, everybody knows that everybody knows that everybody knows, and so on, the values in the payoff matrix.

Byzantine generals problem It is also interesting to note that in message- passing multiagent systems where messages can be lost it is impossible for agents to ever achieve common knowledge about anything. The problem is historically described as the Byzantine generals problem The generals must both attack at the same time in order to defeat the enemy. The only method of communication is by sending a messenger who could be captured by the enemy. R1 R2 B

strategy Normal form game, we define a strategy s to be the set of actions all players take. Pure strategy: one where the agents take a specific action. Mixed strategy: one where the agents take different actions, each with some fixed probabilities.

Rational Game theory assumes that player are rational, which we use as a synonym for selfish. That is, a rational player always acts so as to maximize its utility. (Herbert Simon)

zero-sum games A special type of game are those in which the values in every box of the matrix add up to zero. These games are known as zero-sum games.

Solution Concepts The earliest solution concepts were proposed by Von Neumann. He realized that in any given game an agent could always take the action which maximized the worst possible utility it could get. This is known as the maxmin strategy, or minmax In a game with two agents,i and j, agent i’s maxmin strategy is given by

Minimax theorem For the maxmin strategy we have the minimax theorem which states that a strategy that minimizes the maximum loss, a minmax strategy, can always be found for all two- person zero-sum games. Thus, we know it will work at least for this subset of games.

Lack of Stability The strategy where both players play their maxmin strategy might not be stable in the general case. It can happen that if i knows that j will play its maxmin strategy then i will prefer a strategy different from its maxmin strategy. In the example, the maxmin strategy is (b, d) but if Alice plays d then Bob should play a.

Dominant Strategy S is the dominant strategy for agent if the agent is better off doing s regardless of which strategies the others use. Formally, we say that a pure strategy is dominant for agent if Where represents the strategies of all agents except i

Iterated dominance Dominant strategy can be expanded into the iterated dominance solution in which dominated strategies are eliminated in succession.

Social welfare strategy The social welfare strategy is the one that maximizes the sum of everyone’s payoffs. a social welfare strategy might not be stable. I.e., players may wish to seek better payoffs.

Pareto optimal Strategy A strategy s is said to be Pareto optimal if there is no other strategy such that at least one agent is Pareto optimal better off in and no agent is worse off in than in s. The set of all Pareto strategies for a given problem is formally defined to be the set In Economics the Pareto solution of often referred to as Pareto efficient or, simply, the efficient solution.

Nash equilibrium Strategy s is a Nash equilibrium if for all agents is best strategy given that all the other players will play the strategies in s. The set of all Nash equilibrium strategies for a given game is given by

Nash equilibrium There is no general relation between the Nash equilibrium and the Pareto solution. The most common problem we face when designing multiagent systems is the existence of multiple equilibria.

Famous Games The most famous game of all is the Prisoner’s Dilemma

Battle of the sexes The battle of the sexes is another popular game.

Game of chicken The game of chicken, is also common.

Repeated Games The prisoner’s dilemma the dominant strategy was to defect but both players could have received a higher utility by cooperating. One way to try to get out of this conundrum, and to better simulate real-world interactions, is to let two players play the same game some number of times. This new game is known as the iterated prisoner’s dilemma. One way to analyze repeated games that last for a finite number of periods is to backtrack from the end.

Repeated Games cooperative equilibrium for the iterated prisoner’s dilemma if instead of a fixed known number of interactions there is always a small probability that every interaction will be the last interaction.

Folk theorem The folk theorem tells us that in a repeated game any strategy that is not Pareto dominated by another and where every agent gets a utility that is higher than his maxmin utility is a feasible equilibrium strategy. In these cases, each player knows that he can get his maxmin utility by playing the appropriate action, regardless of what the others do. Thus, no player will be satisfied with less than his maxmin utility.

Games in Extended Form In the early 1980’s Robert Axelrod performed some experiments on the iterated prisoner’s dilemma. He sent out an asking people to submit fortran programs that played the prisoner’s dilemma against each other for 200 rounds. The winner was the one that accumulated the most points. Many entries were submitted.

Games in Extended Form They included the following strategies:  ALL-D: always play defect.  RANDOM: pick action randomly.  TIT-FOR-TAT: cooperate in the first round, then do whatever the other player did last time.  TESTER: defect on the first round. If other player defects then play tit-for-tat. If he cooperated then cooperate for two rounds then defect.  JOSS: play tit-for-tat but 10% of the time defect instead of cooperating. The tit-for-tat strategy won the tournament.

Games in Extended Form In extended form games the players take sequential actions. These games are represented using a tree where the branches at each level correspond to a different player’s actions and the payoffs to all agents are given at the leafs. Extended form games can also represent simultaneous actions by using dotted ellipses

Games in Extended Form

Solution Concepts An extended game has a Nash equilibrium strategy if for all agents it is true that they can’t gain any more utility by playing a strategy different from given that everyone else is playing. Extended form games are a useful way to represent more complex multiagent interactions.

Subgame perfect equilibrium A stronger solution concept is the subgame perfect equilibrium strategy which is defined as one where for all agents and all subgames it is true that can’t gain any more utility by playing a strategy different from.

Finding a Solution There is an extensive literature on centralized algorithms for finding the various equilibrium strategies for a given game. The algorithms involve a complete search using heuristics for pruning. The Gambit software program is an open source implementation of several of these algorithms.