DOCSLIB.ORG

Part 10 Game Theory (Part IV)

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Part 10 Game Theory (Part IV) Part 10 Game theory (part IV) 306 Brief history 1846 (Babbage): machine to play tic-tac-toe 1928 (von Neumann): minimax theorem 1944 (von Neumann & Morgenstern): backward-induction algorithm (produces perfect play) 1950 (Shannon): minimax algorithm (finite horizon, approximate evaluation) 1951 (Turing): program (on paper) for playing chess 1952–7 (Samuel): checkers program, capable of beating its creator 1956 (McCarthy): pruning to allow deeper search 1957 (Bernstein): first complete chess program, on an IBM 704 vacuum-tube computer, could examine about 350 positions/minute 307 Brief history 1967 (Greenblatt): first program to compete in human chess tournaments: 3 wins, 3 draws, 12 losses 1992 (Schaeffer): Chinook won the 1992 US Open checkers tournament 1994 (Schaeffer): Chinook became world checkers champion; Tinsley (human champion) withdrew for health reasons 1997 (Hsu et al): Deep Blue won 6-game chess match against world chess champion Gary Kasparov 2007 (Schaeffer et al, 2007): Checkers solved: with perfect play, it’s a draw. This took 1014 calculations over 18 years 308 Quick review A strategy tells what an agent will do in every possible situation. Strategies may be pure (deterministic) or mixed (probabilistic). Finite perfect-information zero-sum games: I Finite: finitely many agents, actions and states. I Perfect information: every agent knows the current state, all of the actions, and what they do. No simultaneous actions (agents move one-at-a-time). I n Constant-sum: regardless of how the game ends, i=1(ui) = k (for some constant k). For every such game, there’s an equivalent game in which k = 0. Thus constant-sum P games usually are called zero-sum games. Examples: I Deterministic: chess, checkers, go, othello (reversi), connect-four. I Stochastic: backgammon, monopoly For now, we will consider only deterministic games. 309 Quick review Finite two-person perfect-information zero-sum games: I N = Max, Min . { } I Max’s expected utility is uMax(s, t). I From now on, we will just call this u(s, t). I Since G is zero-sum, u (s, t) = u(s, t). Min − I Max wants to maximize u and Min wants to minimize it. 310 Quick review Theorem (Minimax theorem, von Neumann 1928) Let G be a two-person finite zero-sum game. Then there are strategies s∗ and t∗, and a number u∗, called G’s minimax value, such that: I If Min uses t , Max’s expected utility is u , i.e., max u(s, t ) = u . ∗ ≤ ∗ s ∗ ∗ I If Max uses s , Max’s expected utility is u , i.e., min u(s , t) = u . ∗ ≥ ∗ t ∗ ∗ Corollary u(s∗, t∗) = u∗ Corollary If G is a perfect-information game, then there are pure strategies s∗ and t∗ that satisfy the theorem. 311 Strategies on game trees To construct a pure strategy for Max: 1. At each node where it’s Max’s move, choose one branch. 2. At each node where it’s Min’s move, include all branches. Let b be the tree’s branching factor (max. number of children of any node). Let m be the tee’s height (max. depth of any node). m The number of pure strategies for Max is bounded by b 2 . m d e (= bd 2 e if every node of height < m has b children). 312 Strategies on game trees To construct a pure strategy for Min: 1. At each node where it’s Min’s move, choose one branch. 2. At each node where it’s Max’s move, include all branches. m The number of pure strategies for Min is bounded by b 2 . m d e (= bd 2 e if every node of height < m has b children). 313 Finding the best strategy The brute-force algorithm to find Max’s and Min’s best strategies: I Construct the sets S and T of all Max’s and Min’s pure strategies, then choose: s∗ = arg max min u(s, t) s S t T ∈ ∈ t∗ = arg min max u(s, t) t T s S ∈ ∈ Complexity analysis: m m m I Constructs and stores O(b 2 + b 2 ) = O(b 2 ) strategies. m I Each strategy is a tree that has O(b 2 ) nodes. m m m I Thus, space complexity is O(b 2 b 2 ) = O(b ). I Time complexity is slightly worse. But there is an easier way to find best strategies. 314 Backward induction For two-person perfect-information zero-sum games Algorithm function backward-induction(h) returns u(h) if h Z then return u(h) ∈ if ρ(h) = Max then return maxa A (backward-induction(σ(h, a)) ∈ else return mina A (backward-induction(σ(h, a)) ∈ To get action to be performed at h, return the action (arg max or arg min). Complexity analysis: I Space complexity: O(bm). I Time complexity: O(bm). For example, Chess: I b 35 and m 100 (“reasonable” games); ≈ ≈ I 35100 10150 nodes; ≈ I number of particles in the universe: 1087; ≈ I Impossible to explore them all! 315 Minimax algorithm Modified version of the backward induction: I ` is an upper bound on the search depth. I e(h) is a static evaluation function that returns an estimate of u∗(h). I Whenever we reach a non-terminal node h of depth `, return e(h). If ` = , then function e is never called, and Minimax returns u (h). ∞ ∗ Algorithm (Minimax – Shannon, 1950) function minimax(h, `) returns an estimate of u(h) if h Z then return u(h) ∈ if ` = 0 then return e(h) if ρ(h) = Max then return maxa A (minimax(σ(h, a), ` 1)) ∈ − else return mina A (minimax(σ(h, a), ` 1)) ∈ − 316 Evaluation functions e(h) is often a weighted sum of features. E.g., in chess: 1(#white pawns #black pawns) + 3(#white knights #black nights) + ... − − The exact values of e(h) does not matter. The behaviour of the algorithm is preserved under any monotonic transformation of e. Only the order matters: payoff acts as an ordinal utility. 317 Prunning Backward-induction and minimax both look at nodes that do not need to be examined. Max a 3 ≥ Min b 3 f 2 h 14 5 2 ≤ ≤ ≤ ≤ c d e g ? ? i j k 3 12 8 2 14 5 2 Max never goes to node f, because Max gets a higher utility by going to node b. Since f is worse than b, it cannot affect minimax value of node a. Does not know whether h is better or worse than a. Still do not know whether h is better or worse than a. h is worse than a. 318 Alpha-beta algorithm Algorithm function alpha-beta(h, `, α, β) returns an estimate of u(h) if h Z then return u(h) ∈ if ` = 0 then return e(h) if ρ(h) = Max then v ← −∞ forall a χ(h) do ∈ v max(v, alpha-beta(σ(h, a), ` 1, α, β) ← − if v β then return v ≥ α max(α, v) ← else v ← ∞ forall a χ(h) do ∈ v min(v, alpha-beta(σ(h, a), ` 1, α, β) ← − if v α then return v ≤ β min(β, v) ← return v 319 Example Max a (7, ,) ) −∞∞∞ 87 Min b c (7, 8)) ∞ 7 8 Max d (7, ) m ∞ 85 9 Min e (7, ) j (7, 8)) ∞ ∞ 5 α-cutoff 8 Max f (7, ) i k (7, ) l (7, 8) ∞ ∞ 5 8 9 β-cutoff g h 5 3 0 8 9 − 320 Properties of Alpha-beta prunning Alpha-beta is a simple example of reasoning about which computations are relevant (a form of metareasoning). if α minimax(h, `) β, then alpha-beta(h, `, α, β) returns minimax(h, `). ≤ ≤ if minimax(h, `) α, then alpha-beta(h, `, α, β) returns a value α. ≤ ≤ if minimax(h, `) β, then alpha-beta(h, `, α, β) returns a value β. ≥ ≥ Consequently: I If α = and β = , then alpha-beta(h, `, α, β) returns minimax(h, `) −∞ ∞ I If α = , β = , and ` = , then alpha-beta(h, `, α, β) returns u (u). −∞ ∞ ∞ ∗ 321 Properties of Alpha-beta prunning Good move ordering can enable prunning more nodes. The best case is if: I at nodes where it’s Max’s move, children are largest-value first I at nodes where it’s Min’s move, children are smallest-value first h In this case, Alpha-beta’s time complexity is O(b 2 ), which doubles the solvable depth. Worst case is the reverse: I at nodes where it’s Max’s move, children are smallest-value first I at nodes where it’s Min’s move, children are largest-value first In this case, Alpha-beta will visit every node of depth `. ≤ Hence time complexity is the same as minimax: O(bm) 322 Discussion Deeper lookahead (i.e., larger depth bound `) usually gives better decisions. Exceptions do exist: I “Pathological” games in which deeper lookahead gives worse decisions. I But such games are rare. Suppose we have 100 seconds, and we can explore 104 nodes/second: 6 8 I Thus, we can explore 10 35 2 nodes per move. ≈ I Alpha-beta reaches depth 8. I Pretty good chess program. Some modifications that can improve further the accuracy or computation time: I node ordering (see next slide) I quiescence search I biasing I transposition tables I thinking on the opponent’s time 323 Node ordering Recall that we have the best case if: I at nodes where it’s Max’s move, children are largest first I at nodes where it’s Min’s move, children are smallest first h In the best case time complexity is O(b 2 ), which doubles the solvable depth.
Load more

Recommended publications
  • 1 Sequential Games
    1 Sequential Games We call games where players take turns moving “sequential games”. Sequential games consist of the same elements as normal form games –there are players, rules, outcomes, and payo¤s. However, sequential games have the added element that history of play is now important as players can make decisions conditional on what other players have done. Thus, if two people are playing a game of Chess the second mover is able to observe the …rst mover’s initial move prior to making his initial move. While it is possible to represent sequential games using the strategic (or matrix) form representation of the game it is more instructive at …rst to represent sequential games using a game tree. In addition to the players, actions, outcomes, and payo¤s, the game tree will provide a history of play or a path of play. A very basic example of a sequential game is the Entrant-Incumbent game. The game is described as follows: Consider a game where there is an entrant and an incumbent. The entrant moves …rst and the incumbent observes the entrant’sdecision. The entrant can choose to either enter the market or remain out of the market. If the entrant remains out of the market then the game ends and the entrant receives a payo¤ of 0 while the incumbent receives a payo¤ of 2. If the entrant chooses to enter the market then the incumbent gets to make a choice. The incumbent chooses between …ghting entry or accommodating entry. If the incumbent …ghts the entrant receives a payo¤ of 3 while the incumbent receives a payo¤ of 1.
    [Show full text]
  • Lecture Notes
    GRADUATE GAME THEORY LECTURE NOTES BY OMER TAMUZ California Institute of Technology 2018 Acknowledgments These lecture notes are partially adapted from Osborne and Rubinstein [29], Maschler, Solan and Zamir [23], lecture notes by Federico Echenique, and slides by Daron Acemoglu and Asu Ozdaglar. I am indebted to Seo Young (Silvia) Kim and Zhuofang Li for their help in finding and correcting many errors. Any comments or suggestions are welcome. 2 Contents 1 Extensive form games with perfect information 7 1.1 Tic-Tac-Toe ........................................ 7 1.2 The Sweet Fifteen Game ................................ 7 1.3 Chess ............................................ 7 1.4 Definition of extensive form games with perfect information ........... 10 1.5 The ultimatum game .................................. 10 1.6 Equilibria ......................................... 11 1.7 The centipede game ................................... 11 1.8 Subgames and subgame perfect equilibria ...................... 13 1.9 The dollar auction .................................... 14 1.10 Backward induction, Kuhn’s Theorem and a proof of Zermelo’s Theorem ... 15 2 Strategic form games 17 2.1 Definition ......................................... 17 2.2 Nash equilibria ...................................... 17 2.3 Classical examples .................................... 17 2.4 Dominated strategies .................................. 22 2.5 Repeated elimination of dominated strategies ................... 22 2.6 Dominant strategies ..................................
    [Show full text]
  • 1 Bertrand Model
    ECON 312: Oligopolisitic Competition 1 Industrial Organization Oligopolistic Competition Both the monopoly and the perfectly competitive market structure has in common is that neither has to concern itself with the strategic choices of its competition. In the former, this is trivially true since there isn't any competition. While the latter is so insignificant that the single firm has no effect. In an oligopoly where there is more than one firm, and yet because the number of firms are small, they each have to consider what the other does. Consider the product launch decision, and pricing decision of Apple in relation to the IPOD models. If the features of the models it has in the line up is similar to Creative Technology's, it would have to be concerned with the pricing decision, and the timing of its announcement in relation to that of the other firm. We will now begin the exposition of Oligopolistic Competition. 1 Bertrand Model Firms can compete on several variables, and levels, for example, they can compete based on their choices of prices, quantity, and quality. The most basic and funda- mental competition pertains to pricing choices. The Bertrand Model is examines the interdependence between rivals' decisions in terms of pricing decisions. The assumptions of the model are: 1. 2 firms in the market, i 2 f1; 2g. 2. Goods produced are homogenous, ) products are perfect substitutes. 3. Firms set prices simultaneously. 4. Each firm has the same constant marginal cost of c. What is the equilibrium, or best strategy of each firm? The answer is that both firms will set the same prices, p1 = p2 = p, and that it will be equal to the marginal ECON 312: Oligopolisitic Competition 2 cost, in other words, the perfectly competitive outcome.
    [Show full text]
  • Chapter 16 Oligopoly and Game Theory Oligopoly Oligopoly
    Chapter 16 “Game theory is the study of how people Oligopoly behave in strategic situations. By ‘strategic’ we mean a situation in which each person, when deciding what actions to take, must and consider how others might respond to that action.” Game Theory Oligopoly Oligopoly • “Oligopoly is a market structure in which only a few • “Figuring out the environment” when there are sellers offer similar or identical products.” rival firms in your market, means guessing (or • As we saw last time, oligopoly differs from the two ‘ideal’ inferring) what the rivals are doing and then cases, perfect competition and monopoly. choosing a “best response” • In the ‘ideal’ cases, the firm just has to figure out the environment (prices for the perfectly competitive firm, • This means that firms in oligopoly markets are demand curve for the monopolist) and select output to playing a ‘game’ against each other. maximize profits • To understand how they might act, we need to • An oligopolist, on the other hand, also has to figure out the understand how players play games. environment before computing the best output. • This is the role of Game Theory. Some Concepts We Will Use Strategies • Strategies • Strategies are the choices that a player is allowed • Payoffs to make. • Sequential Games •Examples: • Simultaneous Games – In game trees (sequential games), the players choose paths or branches from roots or nodes. • Best Responses – In matrix games players choose rows or columns • Equilibrium – In market games, players choose prices, or quantities, • Dominated strategies or R and D levels. • Dominant Strategies. – In Blackjack, players choose whether to stay or draw.
    [Show full text]
  • Dynamic Games Under Bounded Rationality
    Munich Personal RePEc Archive Dynamic Games under Bounded Rationality Zhao, Guo Southwest University for Nationalities 8 March 2015 Online at https://mpra.ub.uni-muenchen.de/62688/ MPRA Paper No. 62688, posted 09 Mar 2015 08:52 UTC Dynamic Games under Bounded Rationality By ZHAO GUO I propose a dynamic game model that is consistent with the paradigm of bounded rationality. Its main advantages over the traditional approach based on perfect rationality are that: (1) the strategy space is a chain-complete partially ordered set; (2) the response function is certain order-preserving map on strategy space; (3) the evolution of economic system can be described by the Dynamical System defined by the response function under iteration; (4) the existence of pure-strategy Nash equilibria can be guaranteed by fixed point theorems for ordered structures, rather than topological structures. This preference-response framework liberates economics from the utility concept, and constitutes a marriage of normal-form and extensive-form games. Among the common assumptions of classical existence theorems for competitive equilibrium, one is central. That is, individuals are assumed to have perfect rationality, so as to maximize their utilities (payoffs in game theoretic usage). With perfect rationality and perfect competition, the competitive equilibrium is completely determined, and the equilibrium depends only on their goals and their environments. With perfect rationality and perfect competition, the classical economic theory turns out to be deductive theory that requires almost no contact with empirical data once its assumptions are accepted as axioms (see Simon 1959). Zhao: Southwest University for Nationalities, Chengdu 610041, China (e-mail: [email protected]).
    [Show full text]
  • UC Merced Proceedings of the Annual Meeting of the Cognitive Science Society
    UC Merced Proceedings of the Annual Meeting of the Cognitive Science Society Title Language as a Tool for Thought: The Vocabulary of Games Facilitates Strategic Decision Making Permalink https://escholarship.org/uc/item/2n8028tv Journal Proceedings of the Annual Meeting of the Cognitive Science Society, 28(28) ISSN 1069-7977 Authors Keller, Josh Loewenstein, Jeffrey Publication Date 2006 Peer reviewed eScholarship.org Powered by the California Digital Library University of California Language as a Tool for Thought: The Vocabulary of Games Facilitates Strategic Decision Making Jeffrey Loewenstein ([email protected]) McCombs School of Business, 1 University Station B6300 Austin, TX 78712 USA Josh Keller ([email protected]) McCombs School of Business, 1 University Station B6300 Austin, TX 78712 USA Abstract thereby influence key decision makers to favor their desired policies. More than 60 years ago, the sociologist Mills People in competitive decision-making situations often make (1939) argued that problems are perceived relative to a poor choices because they inadequately understand the vocabulary. For example, what counts as murder seems decision they are to make, particularly because they fail to straightforward, but should differ substantially across consider contingencies such as how their opponent will react vegans, anti-abortion activists, lawyers, soldiers and the to their choice (Tor & Bazerman, 2004). Accordingly it would be useful to have a generally applicable vocabulary to guide (hopefully extinct) ritual practitioner of human sacrifice people towards effective interpretations of decision situations. (Clark, 1998). The vocabulary of games provides one such toolkit. We One role of language is to invoke particular framings or presented 508 participants with words from the vocabulary of interpretations.
    [Show full text]
  • Extensive Form Games and Backward Induction
    Recap Perfect-Information Extensive-Form Games Extensive Form Games and Backward Induction ISCI 330 Lecture 11 February 13, 2007 Extensive Form Games and Backward Induction ISCI 330 Lecture 11, Slide 1 Recap Perfect-Information Extensive-Form Games Lecture Overview Recap Perfect-Information Extensive-Form Games Extensive Form Games and Backward Induction ISCI 330 Lecture 11, Slide 2 I The extensive form is an alternative representation that makes the temporal structure explicit. I Two variants: I perfect information extensive-form games I imperfect-information extensive-form games Recap Perfect-Information Extensive-Form Games Introduction I The normal form game representation does not incorporate any notion of sequence, or time, of the actions of the players Extensive Form Games and Backward Induction ISCI 330 Lecture 11, Slide 3 I Two variants: I perfect information extensive-form games I imperfect-information extensive-form games Recap Perfect-Information Extensive-Form Games Introduction I The normal form game representation does not incorporate any notion of sequence, or time, of the actions of the players I The extensive form is an alternative representation that makes the temporal structure explicit. Extensive Form Games and Backward Induction ISCI 330 Lecture 11, Slide 3 Recap Perfect-Information Extensive-Form Games Introduction I The normal form game representation does not incorporate any notion of sequence, or time, of the actions of the players I The extensive form is an alternative representation that makes the temporal
    [Show full text]
  • Restoring Fun to Game Theory
    Restoring Fun to Game Theory Avinash Dixit Abstract: The author suggests methods for teaching game theory at an introduc- tory level, using interactive games to be played in the classroom or in computer clusters, clips from movies to be screened and discussed, and excerpts from novels and historical books to be read and discussed. JEL codes: A22, C70 Game theory starts with an unfair advantage over most other scientific subjects—it is applicable to numerous interesting and thought-provoking aspects of decision- making in economics, business, politics, social interactions, and indeed to much of everyday life, making it automatically appealing to students. However, too many teachers and textbook authors lose this advantage by treating the subject in such an abstract and formal way that the students’ eyes glaze over. Even the interests of the abstract theorists will be better served if introductory courses are motivated using examples and classroom games that engage the students’ interest and encourage them to go on to more advanced courses. This will create larger audiences for the abstract game theorists; then they can teach students the mathematics and the rigor that are surely important aspects of the subject at the higher levels. Game theory has become a part of the basic framework of economics, along with, or even replacing in many contexts, the traditional supply-demand frame- work in partial and general equilibrium. Therefore economics students should be introduced to game theory right at the beginning of their studies. Teachers of eco- nomics usually introduce game theory by using economics applications; Cournot duopoly is the favorite vehicle.
    [Show full text]
  • MS&E 246: Lecture 10 Repeated Games
    MS&E 246: Lecture 10 Repeated games Ramesh Johari What is a repeated game? A repeated game is: A dynamic game constructed by playing the same game over and over. It is a dynamic game of imperfect information. This lecture • Finitely repeated games • Infinitely repeated games • Trigger strategies • The folk theorem Stage game At each stage, the same game is played: the stage game G. Assume: • G is a simultaneous move game •In G, player i has: •Action set Ai • Payoff Pi(ai, a-i) Finitely repeated games G(K) : G is repeated K times Information sets: All players observe outcome of each stage. What are: strategies? payoffs? equilibria? History and strategies Period t history ht: ht = (a(0), …, a(t-1)) where a(τ) = action profile played at stage τ Strategy si: Choice of stage t action si(ht) ∈ Ai for each history ht i.e. ai(t) = si(ht) Payoffs Assume payoff = sum of stage game payoffs Example: Prisoner’s dilemma Recall the Prisoner’s dilemma: Player 1 defect cooperate defect (1,1) (4,0) Player 2 cooperate (0,4) (2,2) Example: Prisoner’s dilemma Two volunteers Five rounds No communication allowed! Round 1 2 3 4 5 Total Player 1 1 1 1 1 1 5 Player 2 1 1 1 1 1 5 SPNE Suppose aNE is a stage game NE. Any such NE gives a SPNE: NE Player i plays ai at every stage, regardless of history. Question: Are there any other SPNE? SPNE How do we find SPNE of G(K)? Observe: Subgame starting after history ht is identical to G(K - t) SPNE: Unique stage game NE Suppose G has a unique NE aNE Then regardless of period K history hK , last stage has unique NE aNE NE ⇒ At SPNE, si(hK) = ai SPNE: Backward induction At stage K -1, given s-i(·), player i chooses si(hK -1) to maximize: Pi(si(hK -1), s-i(hK -1)) + Pi(s(hK)) payoff at stage K -1 payoff at stage K SPNE: Backward induction At stage K -1, given s-i(·), player i chooses si(hK -1) to maximize: NE Pi(si(hK -1), s-i(hK -1)) + Pi(a ) payoff at stage K -1 payoff at stage K We know: at last stage, aNE is played.
    [Show full text]
  • Alpha-Beta Pruning
    CMSC 474, Introduction to Game Theory Game-tree Search and Pruning Algorithms Mohammad T. Hajiaghayi University of Maryland Finite perfect-information zero-sum games Finite: finitely many agents, actions, states, histories Perfect information: Every agent knows • all of the players’ utility functions • all of the players’ actions and what they do • the history and current state No simultaneous actions – agents move one-at-a-time Constant sum (or zero-sum): Constant k such that regardless of how the game ends, • Σi=1,…,n ui = k For every such game, there’s an equivalent game in which k = 0 Examples Deterministic: chess, checkers go, gomoku reversi (othello) tic-tac-toe, qubic, connect-four mancala (awari, kalah) 9 men’s morris (merelles, morels, mill) Stochastic: backgammon, monopoly, yahtzee, parcheesi, roulette, craps For now, we’ll consider just the deterministic games Outline A brief history of work on this topic Restatement of the Minimax Theorem Game trees The minimax algorithm α-β pruning Resource limits, approximate evaluation Most of this isn’t in the game-theory book For further information, look at the following Russell & Norvig’s Artificial Intelligence: A Modern Approach • There are 3 editions of this book • In the 2nd edition, it’s Chapter 6 Brief History 1846 (Babbage) designed machine to play tic-tac-toe 1928 (von Neumann) minimax theorem 1944 (von Neumann & Morgenstern) backward induction 1950 (Shannon) minimax algorithm (finite-horizon search) 1951 (Turing) program (on paper) for playing chess 1952–7
    [Show full text]
  • Rationality Assumptions of Game Theory and the Backward Induction Paradox
    16 Rationality assumptions of game theory and the backward induction paradox Andrew M Col man Backward induction Backward induction is a form of logical reasoning, based on the technique of mathematical induction, that is applicable to certain sequential games. I shall illustrate it with a simple version of the ancient game of nim. There are 20 matches on the table, and two players take turns in removing either one or two matches, the winner being the player who takes the last one, leaving the other player unable to move. Ernst Zermelo (1912) proved the first major theorem of game theory to the effect that, in games of this type, either one player or the other must have a guaranteed winning strategy. In this case, which player wins, and how? The player who moves first wins. The game is easily solved by backward induction. We begin at the end: if it is my turn to move and there are either one or two matches left, then I can obviously win, because in either case I can immediately take the last match, leaving my opponent unable to move. Therefore, if it is my opponent's move and there are three matches left, I can win, because my opponent must take either one or two and will therefore have to leave either two or one, whereupon I can win immediately. Therefore, if it is my opponent's turn to move and there are six matches left, I can win, because my opponent will have to leave either four or five, and in either case I can then leave three, which as I have shown ensures that I can win.
    [Show full text]
  • Backward Induction, Or SPNE’S
    LECTURE 10: GAMES IN EXTENSIVE FORM Jan Zouhar Games and Decisions Sequential Move Games 2 so far, we have only dealt with simultaneous games (players make the decisions at the same time, or simply without knowing what the action of their opponent is) in most games played for amusement (cards, chess, other board games), players make moves in turns → sequential move games many economic problems are in the sequential form: English auctions price competition: firms repeatedly charge prices executive compensation (contract signed; then executive works) monetary authority and firms (continually observe and learn actions) … formally, we describe these as games in extensive form (representation using a game tree) Games and Decisions Jan Zouhar Class Game: Century Mark 3 rules: played by fixed pairs of players taking turns at each turn, each player chooses a number (integer) between 1 and 10 inclusive this choice is added to sum of all previous choices (initial sum is 0) the first player to take the cumulative sum above 100 (century) loses the game prize: 5 extra points for the final test (!!!) Volunteers? Games and Decisions Jan Zouhar Class Game: Century Mark (cont’d) 4 Analysis of the game what’s the winning strategy? broadly speaking, bring the sum to 89; then your opponent can’t possibly win actually, it’s enough to bring it to 78; then you can make sure to make it 89 later reasoning like this, the winning positions are: 100, 89, 78, 67, 56, 45, 34, 23, 12, 1 → the first mover can guarantee a win! winning strategy: in
    [Show full text]