1 / 42

Game Theory

Game Theory. Jacob Foley. http://www.youtube.com/watch?v=HCinK2PUfyk http://www.youtube.com/watch?v=l0ywiYboCLk. Overview. Introduction and history Total-conflict games Partial-conflict games Three-person voting game. What is Game Theory .

noel
Télécharger la présentation

Game Theory

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Game Theory Jacob Foley

  2. http://www.youtube.com/watch?v=HCinK2PUfyk • http://www.youtube.com/watch?v=l0ywiYboCLk

  3. Overview • Introduction and history • Total-conflict games • Partial-conflict games • Three-person voting game

  4. What is Game Theory • Game- two or more individuals compete to try to control the course of events • Uses mathematical tools to study situations involving both conflict and cooperation

  5. History • The first known discussion of game theory occurred in a letter written by James Waldegrave in 1713 • Theory of Games and Economic Behavior by John von Neumann in 1944 • Eight game theorists have won Nobel prizes in economics

  6. Definitions • Player-maybe be people, organizations or countries • Strategies- course of action they may take based on the options available to them • Outcomes- the consequences of the strategies chosen by the players • Preferences- each player has a perfered outcome

  7. Game theory analyzes the rational choice of strategies • Areas Applied • Bargaining tactics in labor-management disputes • Resource allocation decisions • Military Choices in international crises

  8. What makes it different • Analyzes situations in which there are at least two players • The outcome depends on the choices of all the players • Players can cooperate but it is not necessary

  9. Why is it important? • Provided theoretical foundations in economics • Applied in political science (study of voting, elections, and international relations) • Given insight into understanding the evolution of species and conditions under which animals fight each other for territory

  10. Two-Person Total-Conflict • Location Game • Two Young Entrepreneurs with a new restaurant in the mountains • Lisa likes low elevations • Henry likes higher elevations • Routes A, B, and C run east-west • Highways 1, 2 and 3 run north-south • Henry selects one of the routes • Lisa selects one of the Highways • Selection is made simultaneously

  11. Heights of the intersections

  12. How do they choose • Maximin- the maximum value of the minimum numbers in the row of a table • Minimax- the minimum value of the maximum numbers in the columns of a table • Saddlepoint- the outcome when the row minimum and the column maximum are the same

  13. Solution • In total-conflict games, the value is the best outcome that both players can guarantee • In our example the value is 5 • The value is given by each player choosing their maximin and minimax strategies

  14. Example 2: Restricted-Location • Use the same information from previous problem • However, the county officials outlaw restaurants on Route B and Highway 2

  15. Results • There are no saddlepoints • If both choose their minimax and maximin strategy, we will result in 7 • However, they could try to out think the other which could result in 10 or 2

  16. Duel Game

  17. Flawed Approach • Pitcher- If I choose F I hold the batter down to .300 or less but the batter is likely to guess F which gives him at least .200 and actually .300 • Batter- Because the pitcher will try to surprise me with C, I should guess C. I would then average .500. • Pitcher- But if batter guess C, I should really throw F. Thus leading to an average of .100 for the batter

  18. As we see…we can keep going over and over… • Pure Strategy- Each of the definite courses of action that a player can choose • Mixed Strategy- Course of action is randomly chosen from one of the pure strategies by: • Each pure strategy is assigned some probability, indicating the relative frequency with which that pure strategy will be played • The specific strategy used in any given play of the game can be selected at using some appropriate random device

  19. Expected Value of E • In each of the n payoffs, s1, s2, ……, sn, will occur with the probability p1, p2, ………pn, respectively. • The expected Value E • E=p1s1 +p2s2+………..+ pn*sn • And we assume p1+p2+……+pn=1

  20. Matching Pennies • Two players • Each has a penny • They both show either heads or tails at the same time • If the match, player 1 gets the pennies • If they are not a match, player 2 gets the pennies

  21. Payoff Matrix

  22. Results • H & T are pure strategies for both players • There is no way one player can outguess the other • Each player should use a mixed strategy choosing H half the time and T half the time • For player 1: • E(h)= ½(1) + ½(-1) = 0 • E(t)= ½(-1) + ½(1) =0

  23. Cont. • The expected value for player 2 is the same • This means the game is fair, which means the expected value = 0 and therefore favors neither player when at least one player uses an optimal mixed strategy • If one player does not use the 50-50 strategy the player that does gains an advantage

  24. Another example

  25. Results • Player 1 • E(H) = 5*(p) + (-3)(1-p) = 8p-3 • E(T) = (-3)(p) +(1-p)=-4p +1 • 8p-3=-4p+1 • 12p = 4 • P=1/3 • Therefore, • E(H) = 8(1/3) -3 = E(T) = -4(1/3) + 1 =-1/3 => p=1/3 • So their optimal mixed straigy is (1/3, 2/3) with expected value of 1/3

  26. Cont. • Using same calculations for player 2 we get the same optimal mixed stratigy of (1/3, 2/3) • However, the expected value for player 2 is 1/3 • Therefore, we have a zero-sum game.

  27. Lets go back to the baseball game

  28. What should the pitcher do? • E(f)= (0.3)p + (0.2)(1-p) = 0.1p + 0.2 • E(c)= (0.1)p + 0.5(1-p) = -0.4p + 0.5 • Solution is at the intersection of these two lines • -0.4p + 0.5 = 0.1p + 0.2 • p = 0.6 • Giving E(f)=E(c)=E=0.26 • Thus, the Pitcher should pitch F with p = 3/5 and C with p=2/5 so the batter will not be better than .260

  29. What should the batter do? • E(f)= (0.3)q + (0.1)(1-q) = 0.2q + 0.1 • E(c)= (0.2)q + (0.5)(1-q) = -0.3q + 0.5 • 0.2q + 0.1 = -0.3q + 0.5 • q=0.8 • E(f) = E(c) = E = 0.260 • Therefore, he should guess F with p=4/5 and C with p=1/5 which gives him a batting average of 0.260 • So this gives us an outcome of 0.260

  30. Partial-Conflict Games • These are games in which the sum of payoffs to the players at different outcomes varies • There can be gains by both players if the cooperate but this could be difficult

  31. Prisoners’ Dilemma • Two-person variable-sum game • Shows the workings behind arms races, price wars, and some population problems • In these games, each player benefits from cooperating • There is no reason for them to cooperate without a credible threat of retaliation for not cooperating • Albert Tucker, Princeton mathematician, named the game the Prisoners’ Dilemma in 1950

  32. So the actual game • Two people are accused of a crime • Each person has a choice: • Claim their innocence • Sign a confession accusing the partner of committing the crime • It is in their interest to confess and implicate their partner to receive reduce sentence • However, if both confess, both will be found guilty • As a team, their best interest is to deny having committed the crime

  33. Apply it to the real world army race • Two nations, Red and Blue • A: Arm in preparation for war • D: Disarm or negotiate an arms-control agreement • Rank from best to worse (41)

  34. What should they do? • Red • If Blue selects A- Red receives a payoff of 2 for A and 1 for D, so choose A • If Blue select D- Red receives a payoff of 4 for A and 3 for D, so choose A • Red has a dominate strategy of A • So a rational Red nation will choose A • Similarly, Blue will choose A

  35. Results • If the nations work independently, we get an outcome of (A,A) with payoff of (2,2) • This is a Nash Equilibrium- where no player can benefit by departing by itself from its strategy associated with an outcome • So, each player can corporate, play independent, or defect • Defect dominates cooperate and playing independent for both players • However, defect by both players results in a worse outcome than the mutual-cooperation outcome

  36. Another Example “Chicken” • Two Drivers coming at each other at high speeds

  37. Results • Neither player has a dominate strategy • The Nash Equilibrium are (4,2) and (2,4) • This means that getting the result of (3,3) will be unlikely because each players has an incentive to deviate to get a high payoff

  38. Larger Games • Lets look for a 3x3x3 game • We find the optimal solution by looking at individuals dominant strategy • Reducing it to a 3x3 game and we solve like a 2 person games we have been doing

  39. Example: Truel • A duel with 3 people • Each player has a gun and can either fire or not fire at either of the other players • Goal is to survive 1st and survive with as few other players as possible • http://www.youtube.com/watch?v=rExm2FbY-BE&feature=related

  40. Game Tree

More Related