CHAPTER 14:

GAME THEORY
to accompany
Operations Research: Applications & Algorithms,
4th edition, by Wayne L. Winston

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

What is Game Theory?

It deals with Bargaining.
The whole process can be expressed
Mathematically
Based on Behavior Theory, has a more casual
approach towards study of Human Behavior.
It also considers how people Interact in Groups.
2

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Game Theory Definition

Theory of rational behavior for interactive decision problems. In
a game, several agents strive to maximize their (expected)
utility index by choosing particular courses of action, and each
agent's final utility payoffs depend on the profile of courses of
action chosen by all agents. The interactive situation, specified
by the set of participants, the possible courses of action of each
agent, and the set of all possible utility payoffs, is called a
game; the agents 'playing' a game are called the players.

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Definitions
Definition:ZeroSumGameAgamein
whichthepayoffsfortheplayersalwaysadds
uptozeroiscalledazerosumgame.
Definition:MaximinstrategyIfwe
determinetheleastpossiblepayoffforeach
strategy,andchoosethestrategyforwhichthis
minimumpayoffislargest,wehavethe
maximinstrategy.
4

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

A Further Definition
Definition:Constantsumandnonconstantsum
gameIfthepayoffstoallplayersadduptothe
sameconstant,regardlesswhichstrategiesthey
choose,thenwehaveaconstantsumgame.The
constantmaybezerooranyothernumber,so
zerosumgamesareaclassofconstantsum
games.Ifthepayoffdoesnotadduptoa
constant,butvariesdependingonwhich
strategiesarechosen,thenwehaveanon
constantsumgame.
5

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Game theory: assumptions

(1) Each decision maker has available to him two
or more well-specified choices or sequences of
choices.
(2) Every possible combination of plays available
to the players leads to a well-defined end-state
(win, loss, or draw) that terminates the game.
(3) A specified payoff for each player is associated
with each end-state.
6

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Game theory: assumptions (Cont)

(4) Each decision maker has perfect
knowledge of the game and of his opposition.

(5) All decision makers are rational; that is,

each player, given two alternatives, will select
the one that yields him the greater payoff.

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Two-Person Zero-Sum and Constant-Sum Games

Two-person zero-sum and constant-sum games are
played according to the following basic assumption:
Each player chooses a strategy that enables him/her to
do the best he/she can, given that his/her opponent
knows the strategy he/she is following.
A two-person zero-sum game has a saddle point if and
only if
Max (row minimum) = min (column maximum)
all
all
(1)
rows

columns

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Two-Person Zero-Sum and Constant-Sum Games (Cont)

If a two-person zero-sum or constant-sum game has a saddle
point, the row player should choose any strategy (row)
attaining the maximum on the right side of (1). The column
player should choose any strategy (column) attaining the
minimum on the right side of (1).
In general, we may use the following method to find the
optimal strategies and value of two-person zero-sum or
constant-sum game:
Step 1 Check for a saddle point. If the game has none, go on
to step 2.

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Two-Person Zero-Sum and Constant-Sum Games (Cont)

Step 2 Eliminate any of the row players dominated
strategies. Looking at the reduced matrix (dominated rows
crossed out), eliminate any of the column players
dominated strategies and then those of the row player.
Continue until no more dominated strategies can be found.
Then proceed to step 3.

Step 3 If the game matrix is now 2 x 2, solve the game

graphically. Otherwise, solve by using a linear
programming method.
10

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Zero Sum Games

Game theory assumes that the decision maker and the
opponent are rational, and that they subscribe to the
maximin criterion as the decision rule for selecting
their strategy
This is often reasonable if when the other player is an
opponent out to maximize his/her own gains, e.g.
competitor for the same customers.
Consider:
Player 1 with three strategies S1, S2, and S3 and Player
2 with four strategies OP1, OP2, OP3, and OP4.
11

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Zero Sum Games (Cont)

OP1
S1
Player 1
S2
S3
Column
maxima

12
5
3
12

Player 2
OP2
3
4
0
4
minimax

OP3

OP4

9
6
6
9

8
5
7
8

Row
Minima
3
4
0

maximin

Using the maximin criterion, player 1 records the

row minima and selects the maximum of these (S2)
Player 1s gain is player 2s loss. Player 2 records
the column maxima and select the minimum of these
(OP2).
12

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Zero Sum Games (Cont)

The value 4 achieved by both players is
called the value of the game
The intersection of S2 and OP2 is called a
saddle point. A game with a saddle point is
also called a game with an equilibrium
solution.
At the saddle point, neither player can
improve their payoff by switching strategies
13

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Two-person zero-sum game with a

saddle point
Player #2
Player
#1

1
2
3

Row Domination:
(1)

A
6
9
9

B
5
7
8

C
-4
-2
-3

Payoff Matrix
to Player #1

(2) > (1)

(3) >

Eliminate Row (1)

Column Domination:
14

A > C, B > C
Eliminate Columns A, B

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Reduced Matrix
2
3

C
-2
-3

The best strategy for player #1 is to choose 2.

The best strategy for player #2 is to choose C
This results in a saddle/equilibrium point
which gives us these simple strategies for each
player
15

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Saddle Point
1
2
3

A
6
9
9

B
5
7
8

C
-4
-2
-2
-3

The value shown is the smallest in its row - to

player #2s advantage - and the largest in its
column - to player #1s advantage.
16

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Two-Person Nonconstant-Sum Games

Most game-theoretic models of business
situations are not constant-sum games, because it
is unusual for business competitors to be in total
conflict.
As in two-person zero-sum game, a choice of
strategy by each player is an equilibrium point
if neither player can benefit from a unilateral
change in strategy
17

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

The Prisoners Dilemma

The prisoners dilemma is a universal concept. Theorists
now realize that prisoners dilemmas occur in biology,
psychology, sociology, economics, and law. The prisoners
dilemma is apt to turn up anywhere a conflict of interests
exists -- and the conflict need not be among sentient beings.
Study of the prisoners dilemma has great power for
explaining why animal and human societies are organized as
they are. It is one of the great ideas of the twentieth century,
simple enough for anyone to grasp and of fundamental
importance (...). The prisoners dilemma has become one of
the premier philosophical and scientific issues of our time.
It is tied to our very survival (W. Poundstone,1992, p. 9).
18

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Prisoners Dilemma
Two members of a criminal gang are arrested and
imprisoned.
They are placed under solitary confinement and have no
chance of communicating with each other

The district attorney would like to charge them with

a recent major crime but has insufficient evidence
He has sufficient evidence to convict each of them of a
lesser charge
If he obtains a confession from one or both the criminals,
he can convict either or both on the major charge.
19

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

 The district attorney offers each the chance to turn

states evidence.
If only one prisoner turns states evidence and testifies
against his partner he will go free while the other will
receive a 3 year sentence.
Each prisoner knows the other has the same offer
The catch is that if both turn states evidence, they each
receive a 2 year sentence
If both refuse, each will be imprisoned for 1 year on the
lesser charge
20

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

What would you do if you were Prisoner

A?
Prisoner A
refuses deal
A turns state's
evidence

21

Prisoner B refuses deal

B turns state's evidence

1 year, 1 year

3 years, 0 year

0 years, 3 years

2 years, 2 years

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Two-Person Nonzero-sum Games

In such games, one players gain is not equal to the others
loss. Example: Prisoners dilemma
Prisoner B refuses deal
Prisoner A refuses deal
1 year, 1 year
A turns state's evidence
0 years, 3 years

22

B turns state's evidence

3 years, 0 year
2 years, 2 years

Prisoner A thinks: If the other prisoner refuses the deal then

I am better off turning states evidence. If B turns states
evidence, I am also better off turning states evidence.
Prisoner B thinks similarly.
Because there is no communication and no mutual trust, the
rational prisoners obtain outcomes that are worst off than if
they had cooperated.
Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

General questions in two-person games

Communication?
We have assumed that there is no
communication between the two prisoners. What would happen if
they could communicate?
Repetition? In the Prisoners Dilemma, the two prisoners interact
only once. What would happen if the interaction were repeated?
2- vs. n-person Games? The Prisoners Dilemma is a two-person
game, What would happen if there were many players?
Dominance Reasoning? Compelling as the reasoning is that
leads to the dominant strategy equilibrium may be, it is not the
only way this problem might be reasoned out. Is it really the most
rational answer after all?
23

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Linear Programming and Zero-Sum Games

The value of the game and the optimal strategies
for the row and column players reward matrix
may be found by solving the row players LP and
the column players LP, respectively.

24

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Linear Programming and Zero-Sum Games (Cont)

The dual of the row (column) players LP is the column
(row) players LP. The optimal objective function value
for either the row or the column players LP is the value
of the game to the row player. If the row player departs
from her/his optimal strategy, she/he may receive an
expected reward that is less than the value of the game. If
the column player departs from her/his optimal strategy,
she/he may incur an expected loss that exceeds the value
of the game. Complementary slackness may be used to
simultaneously solve the row and the column players
LPs.
25

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Minimax Rule
Goal of game tree search: to determine one move for Max
player that maximizes the guaranteed payoff for a given
game tree for MAX
Regardless of the moves the MIN will take
The value of each node (Max and MIN) is determined by
(back up from) the values of its children
MAX plays the worst case scenario:
Always assume MIN to take moves to maximize his pay-off
(i.e., to minimize the pay-off of MAX)
For a MAX node, the backed up value is the maximum of the
values associated with its children
For a MIN node, the backed up value is the minimum of the
values associated with its children
26

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Minimax procedure
Create start node as a MAX node with current board
configuration
Expand nodes down to some depth (i.e., ply) of lookahead
in the game.
Apply the evaluation function at each of the leaf nodes
Obtain the back up" values for each of the non-leaf nodes
from its children by Minimax rule until a value is computed
for the root node.
Pick the operator associated with the child node whose
backed up value determined the value at the root as the
move for MAX
27

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Minimax Search
2
1

2
2

Static evaluator
value

This is the move

selected by minimax

MAX
MIN
28

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

N-person games
With a third player, the possibility of a coalition
arises
Within the coalition there are issues of payoff
distribution
Coalitions disintegrate / realignments occur
Solutions for N-person games can be difficult
29

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

N-person games (Cont)

When more than two players are involved, the
structure of a competitive situation may be
summarized by the characteristic function. For each
set of players S, the characteristic function v of a
game gives the amount v (S) that the member of S
can be sure of receiving if they act together and form
a coalition.
Let x x , x ,..., x be a vector such that player i
receives a reward x . We call such a vector a reward
vector. A reward vector x x , x ,..., x is an
i n
imputation if and only if
1

v N xi
i 1

30

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

N-person games (Cont)

xi v i For each i N

The imputation y y , y ,... y dominates

x
S
through a coalition S ( Written y x ) if
1

y
iS

v S

And for all

i S , y xi
i

The core and the Shapley value are two alternative

solution concepts for n-person games. The core of an nperson game is the set of all undominated imputations. An
imputation x x1 , x2 ,..., xn is in the core of an n-person
game if and only if for each subset S of N 1,2,..., n
31

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

N-person games (Cont)

x v S
i

iS

The Shapley value gives a reward x, to the i th player,

where x is given by

x Pn S v S i v S
i

all S for which

i is not in S

Where

P S
n

32

S ! n S 1 !
n!

The last equation implies that player is reward should be

the expected amount that player i adds to the coalition
made up of the players who are present when player i
arrives.
Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.

Discussion/Conclusion
Will Game Theory give us the optimum or
best solution/decision?

33

Copyright 2004 Brooks/Cole, a division of Thomson Learning, Inc.