You are on page 1of 48

Game Theory

Game theory is involved in various conflict and competition. Numerous examples involving conflicts include parlor games, military battles, political campaigns, advertising and marketing campaigns etc.
Game theory is a mathematical theory that deals with the general features of competitive situation. Definition: Game is defined as an activity between two or more persons involving activities by each person according to a set of rules and at the end of which each person receives some benefit or satisfies or suffers loss.

Characteristics of Game Theory The game is classified on the basis of characteristics. The various characters of a game are 1.Chance of strategy: Activities are determined by skill, it is said to be a game of strategy. If the activity are determined by change, it is a game of change. 2. Number of Person: A game is called a n-person game if the number of persons playing is n. The person means an individual or a group aiming at a particular objective.

3. Number of Activities: There may be finite or infinite.

4. Number of Alternations (Choices): A finite game will have finite number of activity, each involving a finite number of alternatives, otherwise the game is said to be infinite. 5. Players information: For each player; part activities to known to the other players either partly, completely or not availability at all. 6. Pay off: It is a quantitative measure of satisfaction a person gets at the end of each play. If the algebraic sum of all the player payoff is zero, the game is said to be zero sum game.

Basic Definitions:
Strategy: Strategy for a given player is a set of rules that specifies which of the available course of action he should make at each play. Pure Strategy: If a player knows exactly what the other player is going to do, the pure strategy is decision rule always to select a particular course of action. Mixed Strategy: Mixed strategy is the course of action taken by the player opting more than one strategy with fixed probabilities.

Two Person, Zero-Sum Game: A game is said to be two person, zero-sum game if only two players play the game and if the losses of one player are equivalent to the gains of the other, so that the sum of their net gains is zero.
Pay-Off Matrix: Suppose the player A has m activities and player B has n activities, then the payoff matrix can be formed by adopting the following rules. (i) Row designation of each matrix are activities available for player-A. (ii)Column designations for each matrix are activities available to player B.

(iii) Cell entry Cij is the payment to player A, is As payoff matrix when A chooses the activity i and B chooses the activity j.
Player B 1 1 Player A 2 C11 C21 2 C12 C22 3 C13 C23 n C1n C2n

Cm1

Cm2

Cmn

Plan ahead and look back

If you have a Dominating strategy, use it


Use strategy 1

Eliminate strategy 2 as its dominated by strategy 1

Eliminate any Dominated strategy

Maximin & Minimax Equilibrium


Minimax - to minimize the maximum loss (defensive) Maximin - to maximize the minimum gain (offensive) Minimax = Maximin

MIN MAX (MAX MIN) Principal The MinMax criteria of optimality states that if player list the worst possible outcomes of all the potential strategies, he will choose that strategy to be the most suitable for him which corresponds to the best of these worst outcomes. 1. Find the Max-Min and Min-Max for the following.
1 A 1 2 3 -3 2 5 B 2 -2 0 -2 3 6 2 -4

B 1 2 3

Col. Max

1 -3 -3 2 5 5

2 -2 0 0 -2 0 0

3 6 2 -4 -4 6

Row Min -3 0 -4

As Max(Min) = Bs Min(Max) Max(-3, 0, -4) = Min (5, 0, 6) = 0 Here there is a saddle point The value of the game is 0

2. Find the value of the game.

B 1 A 1 2 3 1 1 0 2 2 0 1 3 4 5 -1

B 1 1 2 2 3 4 Row Min 1

2 3

1 0
1

0 1
2

5 -1
5

0 -1

Col Max

Max (Min) =1
Max Min = Min max.

Min(Max) = 1

Hence the value of the game is 1.

Problem with Mixed Strategy: Let player A choose a strategy Head H with a probability x and the strategy Tail T with a probability (1-x). B
H T -1 0

H T

2 -1

-1 x -1 (1-x)

In this problem, Max (MIN) Min(Max). Hence it does have a saddle point. If the player B plays H all the time, As expected gain will be

E(A,H) = x.2 + (1-x) (-1) = 3x-1


If player B plays T all the time, As expected gain will be E(A, L) = x(-1) + 0 (1-x) = -x

E(A, L) = E(A, H) = E(A) say


Then 3x -1 = -x

Therefore = .
Therefore the best strategy for the player A is to play H and T with probability and respectively.

E(A) = Value of he game = *2 + *(-1) = -1/4

Dominance Principle It is always convenient to deal with smaller payoff matrix. To reduce the matrix, we have to consider inferior and superior strategies. Considering two n-tuples a (a1,a2,a3, . . . , an) and b = (b1,b2,. . .,bn), if ai >= bi for all i=1,2,n, then for player a strategy corresponding to A is said to be superior to the strategy corresponding to b. The player B, the above situation will be reversed because player As gain matrix is player Bs loss matrix.

Summary of Dominance Rules:


Rule-1: If each element in one row, say rth of the payoff matrix [cij] is less than or equal to the corresponding element in the other row, say sth, then player A will never choose the rth strategy. rth row is said to be dominated by the sth row. Hence rth row can be eliminated. Rule-2: If each element in one column say Cr, is greater than or equal to the corresponding element in the other column, say, Cs, then the player B will never use the strategy corresponding to column Cr. The column Cs dominates column Cr.

Rule-3: Dominance need not be based on the superiority of pure strategies only. A Given strategy can be dominated if it is inferior to an average of two or more other pure strategies. If the dominance holds strictly, then the value of optimal strategies do coincide and then the dominance does not hold strictly, then optimal strategies may not coincide.

Example:
B II 7

III 2

Min 1

I 1 A II 6 III 5
Max 6

2
2 7

7
6 7

2
2

Max (Min) Min (Max). Hence saddle point does not exist. Using the principle of dominance, Player A will never play strategy III as II dominates. Hence eliminate strategy III.

I I 1 6

B II 7 2

III 2 7

A
II

Player B will never play strategy III as strategy I dominates over III.
B 1 A 6 2 7

Let player A play strategy I with probability x and II with probability (1-x).
E(A, I) = x + 6(1-x) = -5x + 6 E(A, II) = 7x + 2(1-x) = 5x +2 E(A) = E(A, I) = E(A, II) -5x +6 = 5x +2 Therefore x = 2/5 (1-x) = 3/5. Value of the game is 4.

1. Use the relation of dominance to solve the rectangular game whose payoff matrix to A is given in B the table. 1 2 3 4 5 6 1 0 0 0 0 0 0 2 4 2 0 2 1 1 A 3 4 3 1 3 2 2 4 4 3 7 -5 1 2 5 4 3 4 -1 2 2 6 4 3 3 -2 2 2

Consider the 4 x 4 game played by Players A and B and solve it optimally.


1 6 2 2 5 2 2 -1 3 2 3 4 1 3 6 4 8 12 9 10

1 2 3 4

Solution: The optimal solution of the problem is: A = (1/5, 0, 4/5, 0) B = (1/5, 4/5, 0, 0) Value of the game = 2.8

Solve the following 3 x5 game using dominance property.

1
1 2 3 2 3 4

2
5 3 4

3
10 6 8

4
7 6 12

5
2 4 1

A (0, , ) B (3/4, 0,0,0,1/4) Value of the gem = 3.25

Graphical Method for 2xn or m x 2 Games The payoff matrix of the 2 x n game will contain 2 rows and n columns, whereas the payoff matrix of the m x 2 game contains m rows and 2 columns. If there is no saddle point for these games, once can solve them using graphical method. Algorithm for 2 x n game 1. Reduce the size of the payoff matrix of player A by applying the dominance property. If it exists. 2. Let, x be the probability of selective 1 by player A and (1-x) be the probability of selection of

3.

4.

5.

6.

Alternative 2 by player A. Derive the expected gain function of Player A with respect to each of the alternatives of Player B. For each of the gain functions which are derived in step 2, find the value of the gain, when x is equal to 0 as well as 1. Plot the gain functions on a graph by assuming a suitable scale. Keep x on X-axis and the gain on Y-axis Since the Player A is a maximum player, find the highest intersection point in the lower boundary of the graph. Let it be the maximum point. If the number of lines passing through the maximin point is only two, form a 2x2 payhoff

matrix from the original problem by retaining only the columns corresponding to these two lines and go to step 8, other wise go to step 7.
7. Identify any two lines with opposite slopes passing through that point. Then form a 2 x 2 payoff matrix from the original problem by retaining only the columns corresponding to those two lines which are having opposite slopes. 8. Solve the 2x2 game using oddments and find the strategies for Players A and B and also the value of the game.

Consider the payoff matrix of Player A and solve it optimally using graphical method.
Player B Player A 1 2 1 3 -1 2 0 5 3 6 -2 4 -1 2 5 7 1

Row minimum = (-1, -2) Maximum of Row minimum = -1. Column maximum (3,5,6,2,7) Minimum of Column minimum = 2 Max (min) is not equal to Min (max). Hence no saddle.

Now checking dominance property: If the payoff matrix as shown in table, column 5 is dominated by column 3. similarly, column 2 is dominated by column 4. Hence delete column 2 and column 4. The resultant payoff matrix after deleting columns 2 and 5. Player B
1 3 -1 3 6 -2 4 -1 2

Player A

1 2

X (1-x)

Let player A play strategy-1 with probability x and strategy 2 with probability (1-x).

For each of the pure strategies available for player B, the expected payoff for the player A, is as follows:
B's Pure A's Expected Payoff Strategies B1 3. x + (1-x) (-1) = 4x -1 B3 6 x + (-2) (1-x) = 8x -2 B4 (-1) x + (2 (1-x) = -3x +2 1. Construct two vertical axes, axis 1 at the point x1=0 and axis 2 at the point x1 = 1. 2. Represent the payoffs v2j, j=1,2,3n on axis 1 and payoff v1j, j = 1,2,n on axis 2.

3. Join the point representing v1j on axis 1. The resulting straightline is the expected payoff line Ej(xi), j=1,2,n.
4. Mark the lowest boundary of the lines Ej(x1) so plotted, by thick line segments. The highest point on this lowest boundary gives the maximum point P and identifies the two critical moves of player B. If there are more than two lines passing through the maximum point P, there are ties for the optimum mixed strategies for player B.

Maximum point

X=0

X=1

1 2

3 4 6 -1 -2 2

Hence solving the above two strategies we get 6x 2(1-x) = v -x + 2(1-x) =v The strategy played by player 4x-1 = -3x +2 7x = 3 X = 3/7 (1-x) = 4/7 The value of the game is 9/7 A is (3/7, 4/7) Value of the game is 9/7

Let player B play strategy 3 with probability y and strategy 4 with probability (1-y) E(B, 1) = 6y + (-1) (1-y) = 5y -1 E(B, 2) = -2y + (1-y) = -3y +1 5y-1 = -3y + 1 8y = 2 y = The strategy played by player B = ( 0, 0, , ,0) The strategy played by player A is (3/7, 4/7) The strategy played by player B = ( 0, 0, , ,0) Value of the game is 9/7

You might also like