Professional Documents
Culture Documents
Game Theory: By: Purvi R. Chaudhary
Game Theory: By: Purvi R. Chaudhary
By : Purvi R. Chaudhary
U07CO241 B Tech.(Computer Engg.) SVNIT,Surat
Used in social science(economics) Biology (evolutionary biology,ecology) Engineering,computer science,political science,international relations,philosophy
THEORY OF GAMES
The theory of games concerned with decision-making in situations where two or more rational players are involved in conditions of competition and conflicting interests. Their actions influenced each other. The theory of games determine what is the best course of action for them.
There may be two or more than two persons playing a game. If the number of players is 2 then the game is 2 person game and if the number of players is n then it is n-person game.
A game in which the gain of one player is equal to loss of the other player, is called a zero sum game. Here the sum of gains or losses of both (all) players is zero. But if sum of gains or losses is not equal to zero, we call it as non-zero sum game.
A game in which there are only two players and the gains made by one is equal to the loss incurred by others called 2-person zero-sum game.
For example : Suppose two companies COCACOLA and PEPSI, are struggling for a larger share in the market. Now any market share gained by COLA must be the lost share of PEPSI and therefore, the sum of the gains and losses equals zero. If COLA gain 60% market share then the lost share of PEPSI will be 60%. (+60%) + (-60%) = 0
PAY-OFF
The outcomes of a game due to adopting the different courses of action by the competing players in the form of gains or losses for each of the player is known as pay-off. PAY-OFF MATRIX : To show the pay-offs of a game through a rectangular matrix is called pay-off matrix.
STRATEGY
The pre-determined rule by which a player decides his courses of action from his own list of courses of action during the game called strategy. There are four types of strategies: 1) PURE STRATEGY: It is the pre-determined course of action to be employed by the player. The player knows it in advance. 2) MIXED STRATEGY: In it the player decides his course of action in accordance with the some fixed probability distribution. Probabilities are associated with each course of action and the selection is done as per these probabilities. In mixed strategy the opponent cannot be sure of the course of action to be taken of any particular type.
3) DOMINATED STRATEGY: Sometimes in a game we come across a fact that one or more of pure strategies of a player are inferior to at least one of the remaining strategies. In such case this inferior strategy is never used. This inferior pure strategy is dominated by a superior pure strategy. 4) OPTIMAL STRATEGY: A course of action which puts the player in the most preferred position, irrespective of the strategy of his competitors is called an optimal strategy.
The solution to a game calls for determining an optimal strategy for each of the players and the value of game. Each player lists the possible outcomes from his action and selects the best action to achieve as per his objectives. This criteria of optimality is expressed as Maximin for the maximizing player and Minimax for the minimizing player.
MAXIMIN CRITERIA: The maximizing player lists his minimum gains from each strategy and selects the strategy which gives him maximum out of these minimum gains. MINIMAX CRITERIA: The minimizing player lists his maximum loss from each strategy and selects the strategy which gives him minimum loss out of these maximum losses.
DECISION MAKING
UNDER CERTAINTY
UNDER UNCERTAINTY
UNDER
RISK
NON-COMPETITIVE SITUATIONS
COMPETITIVE SITUATIONS
PURE STRATEGY
MIXED STRATEGY
PURE STRATEGY GAME OR GAME WITH SADDLE POINT: B B1 B2 B3 row minima A A1 1 2 2 1 Maximin A2 0 -4 -1 -4 A3 -2 3 -2 -2
1 3 Minimax 2
Column Maxima
2)
A
x
A1 A2
B1 a11 a21
B2 a12 a22
y a22 a12 a11 a22 a21 a12
DOMINANCE METHOD:
For the row side, the strongest row dominates the weakest row. For the column side, the weakest column dominates the strongest column.
(1)
(1)
,if B consistently plays strategy 1 ,if B consistently plays strategy 2 ,if B consistently plays strategy 3
3 p1 2 p2 4 p1 3 p2 p1 + 3 p2
p1=0, p2=1
p1=1, p2=0
p1=5/7
p1=0, p2=1
Formulation for A- Max u u <= 3p1 - 2p2 u <=4p1 - 3p2 u <= p1 + 3p2 ,u is unrestricted(means p1, p2 >= 0 can be positive or negative) p1 + p2 =1 u = Min Profit ( minimum of the 3 profits )
P ( Primal )
LP solution
For DUAL of A,rewrite Primal P
Dual variable for each constraint y1 y2 y3 W u - 3p1 + 2p2 <= 0 u - 4p1 + 3p2 <= 0 u - p1 - 3p2 <= 0,u is unrestricted p1 + p2 = 1 p1, p2 >= 0
Dual to this problem 0.y1 + 0.y2 + 0.y3 + 1.W = W , so dual will be to minimize W So, 1st contraint :- y1 + y2 + y3 + 0.W = 1 -3y1 - 4y2 - y3 + W >= 0 2y1 + 3y2 - 3y3 + W >=0 y1,y2,y3 >= 0, W is unrestricted
LP solution
So, Dual of A
W >= 3y1 + 4y2 + y3 W >= 2y1 -3y2 + 3y3 y1 , y2 , y2 >= 0 , y1 + y2 + y3 = 1 Formulation for BMin v (v=Max Loss that A v >= 3q1 + 4q2 +q3 afflicted on B) v >=-2q1 3q2 + 3q2 q1 , q2 ,q3 >= 0 , q1 + q2 +q3 = 1 So , primal-Dual relationship Now , use duality principle to get q1,q2,q3
LP solution
In Formulation of A putting p1=5/7 , p2= 2/7 to give u=11/7 u <= 3p1 - 2p2 (gives 11/7 -> solved as an equation) u <=4p1 - 3p2 u <= p1 + 3p2 (gives 11/7 -> solved as an equation) In Formulation of B v >= 3q1 + q3 (writing q2 as 0,as its non-basic) v >= -2q1 + 3q3 Solving we get q1 = 2/7 , q2 = 0 , q3= 5/7 so we get v = 11/7
Nash equilibrium
Simple insight : cannot predict the result of the choices of multiple decision makers if we analyze those decisions in isolation. must ask what each player would do, taking into account the decision-making of the others . Each player is assumed to know equilibrium strategies of the other player strategy profile x* S is a Nash equilibrium (NE) if no unilateral deviation in strategy by any single player is profitable for that player, that is
Si is the strategy set for player i, S=S1 X S2 ... X Sn is the set of strategy profiles and f=(f1(x), ..., fn(x)) is the payoff function.x i be a strategy profile of all players except for player i.
Prisoner's dilemma
The prisoner's dilemma is a fundamental problem in game theory prisoner's dilemma forms a non-zero-sum game in which two players may each either cooperate with or defect from (betray) the other player
Prisoner's dilemma
If player 2 Cooperate , player 1 is better off
by defecting If player 2 Defects , Player 1 is better off by Defecting So, obvious Player 1 will defect in this game
If player 1 Cooperate , player 2 is better off by defecting If player 1 Defects , Player 2 is better off by Defecting So, obvious Player 2 will defect in this game
We say, Defect strategy strictly dominates Cooperate strategy Rational players never play strictly Dominated strategy by rule Thus , equilibrium is defect-defect ,Not cooperate-cooperate as their tempatation is for over the other prisoner by defecting them so defect-defect is the only stable outcome