M.

Sc Mini Project-1(Report)

on
Bayesian Games

By
Kulamani Sahoo
Enrollment id:2023MAM033

project submitted to
IIEST,Shibpur

under supervisuion and guidance of

Professor Dr.Ujjal Debnath

In partial fulfillment for the award of the degree of

M.Sc in Applied Mathematics

INDIAN INSTITUTE OF ENGINEERING SCIENCE

AND TECHNOLOGY,SHIBPUR
HOWRAH,WEST BENGAL

1
 CERTIFICATE

This is to certify that the work contained in this project report entitled
“BAYESIAN GAMES” submitted by Kulamani Sahoo
(Enrollment ID: 2023MAM033) to Indian Institute of En-
gineering Science and Technology, Shibpur towards partial re-
quirement for completion of the 1st Semester of Masters of Science in
Applied Mathematics has been carried out by him under my supervision.
The result of this project work or any part there of has not been submit-
ted elsewhere for the award of any degree or diploma.

Project Supervisor
(Prof.Dr. Ujjal Debnath)

External Examiner

DEPARTMENT OF MATHEMATICS
(HOD, Prof. Dr.PRITHA DAS)

2
 ACKNOWLEDGEMENT

I would like to take this opportunity to express my profound gratitude

to my project supervisor Prof. Dr.Ujjjal Debnath for her exem-
plary guidance and constant encouragement throughout the course of
this project.I am grateful to IIEST,Shibpur for providing me the op-
portunity to complete my dissertation.In this project I have tried to give
a brief overview of Bayesian Games.I have benefited greatly from ad-
vantages of the Latex software typing system.It helped me a lot for writing
my project easily.

I would like to extend my deepest gratitude and thanks to my parents

with out whose constant support and encouragement,this project could
never has been completed. I would also like to thank my friends for their
valuable suggestions and support for this project.

Kulamani Sahoo
Contents

1 Some basic defination of Game theory 6

1.1 Game: . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Player: . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Strategy: . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Pure Strategy: . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Mixed Strategy: . . . . . . . . . . . . . . . . . . . . . . 6
1.6 Optimum Strategy: . . . . . . . . . . . . . . . . . . . . 6
1.7 Two Person Zero-Sum Game: . . . . . . . . . . . . . . . 6
1.8 Fair Game: . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.9 Payoff Matrix: . . . . . . . . . . . . . . . . . . . . . . . 7
1.10 Repeated Game: . . . . . . . . . . . . . . . . . . . . . . 7
1.11 Nash Equilibrium: . . . . . . . . . . . . . . . . . . . . . 7
1.12 Bayesian Game: . . . . . . . . . . . . . . . . . . . . . . 7

2 MOTIVATIONAL EXAMPLE: 8

3 GENERAL DEFINITION 13

4 Cournot’s duopoly game with imperfect information: 15

4.1 Incomplete Information about Costs: . . . . . . . . . . . 16
4.2 Demand Uncertainty: . . . . . . . . . . . . . . . . . . . 16
4.3 Strategic Information Gathering: . . . . . . . . . . . . . 16
4.4 Learning and Adaptation: . . . . . . . . . . . . . . . . . 17
4.5 Information Asymmetry: . . . . . . . . . . . . . . . . . . 17
4.6 Example: . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5 CONCLUSION 20

6 REFERENCE: 21

4
 INTRODUCTION

An assumption underlying the notion of Nash equilibrium is that each

player holds the correct belief about the other players’ actions. To do so,
a player must know the game she is playing; in particular, she must know
the other players’ preferences. In many situations the participants are
not perfectly informed about their opponents’ characteristics: bargainers
may not know each others’ valuations of the object of negotiation, firms
may not know each others’ cost functions, combatants may not know each
others’ strengths, and jurors may not know their ’colleagues’ interpreta-
tions of the evidence in a trial. In some situations, a participant may be
well informed about her opponents’ characteristics, but may not know
how well these opponents are informed about her own characteristics. In
this chapter I describe the model of a “Bayesian game”, which general-
izes the notion of a strategic game to allows us to analyze any situation
in which each player is imperfectly informed about some aspect of her
environment relevant to her choice of an action.

5
1 Some basic defination of Game theory

1.1 Game:

Any set of circumstances that has result dependent on the actions of two
or more decision makes(players) is known as game.

1.2 Player:

ctionplayer: A competator in a game known as player.

1.3 Strategy:

A set of alternative courses of action available to player based on advance

knowledge is known as strategy.

1.4 Pure Strategy:

If a player select the same strategy every time,then it is called as pure

strategy.

1.5 Mixed Strategy:

If the player selects his course of action in accordance with some fixed
probability,then it is called mixed strategy.

1.6 Optimum Strategy:

A course of action,which puts the player in the most preferred portion is

known as optimum strategy.

1.7 Two Person Zero-Sum Game:

When only two player are involved in the game and the gain of the player
is equal to loss of the other then the game is known as two person zero
sum-game.
6
1.8 Fair Game:

If the value of the game is zero then it is called as fair game.

1.9 Payoff Matrix:

When two players are playing game they have choices.Choice of one player
becomes the row and choice of other player becomes the column.The
matrix formed by these rows and columns is known as payoff matrix.

1.10 Repeated Game:

A game in which actions are taken and payoffs are received again and
again between two same players is known as repeated games.

1.11 Nash Equilibrium:

Nash equilibrium is a component of game theory that asserts that a player

will continue with his chosen strategy while knowing his opponent’s strat-
egy as they have no incentive to change the rules.

1.12 Bayesian Game:

In shortly we can say that these games in which there is uncertainty

regarding payoff of others are known as Bayesian Game.

7
2 MOTIVATIONAL EXAMPLE:

Example(1): (Variant of BoS with imperfect information) Consider

a variant of the situation modeled by BoS in which player 1 is unsure
whether player 2 prefers to go out with her or prefers to avoid her, whereas
player 2, as before, knows player 1’s preferences. Specifically, suppose
player 1 thinks that with probability 21 player 2 wants to got out with
her, and with probability 12 player 2 wants to avoid her. (Presumably
this assessment comes from player 1’s experience: half of the time she
is involved in this situation she faces a player who wants to go out with
her, and half of the time she faces a player who wants to avoid her.)
That is, player 1 thinks that with probability 21 she is playing the game
on the left of Figure and with probability 12 she is playing the game on
the right. Because probabilities are involved, an analysis of the situation
requires us to know the players’ preferences over lotteries, even if we are
interested only in pure strategy equilibria; thus the numbers in the tables
are Bernoulli payoffs.

8
 In figure A variant of BoS in which player 1 is unsure whether player
2 wants to meet her or to avoid her. The frame labeled 2 enclosing each
table indicates that player 2 knows the relevant table. The frame labeled
1 enclosing both tables indicates that player 1 does not know the relevant
table; the probabilities she assigns to the two tables are printed on the
frame.
We can think of there being two states, one in which the players’ Bernoulli
payoffs are given in the left table and one in which these payoffs are given
in the right table. Player 2 knows the state—she knows whether she
wishes to meet or avoid player 2—whereas player 1 does not; player 1
assigns probability 12 to each state. The notion of Nash equilibrium for a
strategic game models a steady state in which each player’s beliefs about
the other players’ actions are correct, and each player acts optimally, given
her beliefs. We wish to generalize this notion to the current situation.
From player 1’s point of view, player 2 has two possible types, one whose
preferences are given in the left table of Figure , and one whose preferences
are given in the right table. Player 1 does not know player 2’s type, so to
choose an action rationally she needs to form a belief about the action of
each type. Given these beliefs and her belief about the likelihood of each
type, she can calculate her expected payoff to each of her actions. For
example, if she thinks that the type who wishes to meet her will choose B
and the type who wishes to avoid her will choose S, then she thinks that
B will yield her a payoff of 2 with probability 21 and a payoff of 0 with
probability 12 , so that her expected payoff is 12 .2+ 12 .0 = 1, and S will yield
her an expected payoff of 12 .0 + 12 .1 = 12 .Similar calculations for the other
combinations of actions for the two types of player 2 yield the expected
payoffs in Figure 2. Each column of the table is a pair of actions for the
two types of player 2, the first member of each pair being the action of
the type who wishes to meet player 1 and the second member being the
action of the type who wishes to avoid player 1.

9
 For this situation we define a pure strategy Nash equilibrium to be a
triple of actions, one for player 1 and one for each type of player 2, with
the property that
• the action of player 1 is optimal, given the actions of the two types of
player 2 (and player 1’s belief about the state)
•the action of each type of player2 is optimal, given the action of player1.

That is, we treat the two types of player 2 as separate players, and
analyze the situation as a three-player strategic game in which player 1’s
payoffs as a function of the actions of the two other players (i.e. the two
types of player 2) are given in Figure 2, and the payoff of each type of
player 2 is independent of the actions of the other type and depends on
the action of player 1 as given in the tables in Figure 1 (the left table for
the type who wishes to meet player 1, and the right table for the type
who wishes to avoid player 1). In a Nash equilibrium, player 1’s action
is a best response in Figure 2 to the pair of actions of the two types of
player 2, the action of the type of player 2 who wishes to meet player 1
is a best response in the left table of Figure 1 to the action of player 1,
and the action of the type of player 2 who wishes to avoid player 1 is a
best response in the right table of Figure 1 to the action of player 1. Why
should player 2, who knows whether she wants to meet or avoid player 1,
have to plan what to do in both cases? She does not have to do so! But
we, as analysts, need to consider what she does in both cases, because
player 1, who does not know player 2’s type, needs to think about the
10
action each type would take; we would like to impose the condition that
player 1’s beliefs are correct, in the sense that for each type of player 2
they specify a best response to player 1’s equilibrium action. I claim that
(B, (B, S)), where the first component is the action of player 1 and the
other component is the pair of actions of the two types of player 2, is a
Nash equilibrium. Given that the actions of the two types of player 2 are
(B, S), player 1’s action B is optimal, from Figure 2; given that player
1 chooses B, B is optimal for the type who wishes to meet player 2 and
S is optimal for the type who wishes to avoid player 2, from Figure 1.
Suppose that in fact player 2 wishes to meet player 1. Then we interpret
the equilibrium as follows. Both player 1 and player 2 chooses B; player
1, who does not know if player 2 wants to meet her or avoid her believes
that if player 2 wishes to meet her she will choose B, and if she wishes to
avoid her she will choose S.
Example: (Expected payoffs in a variant of BoS with imperfect infor-
mation) Construct tables like the one in Figure 3 for type n1 of player 1,
and for types y2 and n2 of player 2. I claim that ((B, B), (B, S)) and ((S,
B), (S, S)) are Nash equilibria of the game, where in each case the first
component gives the actions of the two types of player1.

Figure 3 The expected payoffs of type y1 of player 1 . Each row cor-

responds to a pair of actions for the two types of player 2; the action of
type y2 is listed first, that of type n2 second. and the second component
gives the actions of the two types of player 2. Using Figure 3 you may
11
verify that B is a best response of type y1 of player 1 to the pair (B,
S) of actions of player 2, and S is a best response to the pair of actions
(S, S). You may use your answer to verify that in each of the claimed
Nash equilibria the action of type n1 of player 1 and the action of each
type of player 2 is a best response to the other players’ actions. In each
of these examples a Nash equilibrium is a list of actions, one for each
type of each player, such that the action of each type of each player is a
best response to the actions of all the types of the other player, given the
player’s beliefs about the state after she observes her signal. The actions
planned by the various types of player i are not relevant to the decision
problem of any type of player i, but there is no harm in taking them, as
well as the actions of the types of the other player, as given when player
i is choosing an action. Thus we may define a Nash equilibrium in each
example to be a Nash equilibrium of the strategic game in which the set
of players is the set of all types of all players in the original situation.

In the next section I define the general notion of a Bayesian game,

and the notion of Nash equilibrium in such a game. These definitions
require significant theoretical development. If you find the theory in the
next section heavy-going, you may be able to skim the section and then
study the subsequent illustrations, relying on the intuition developed in
the examples in this section, and returning to the theory only as necessary
for clarification

12
 3 GENERAL DEFINITION

A Bayesian game consists of

• a set of players

• a set of states and for each player

• a set of actions

• a set of signals that she may receive and a signal function that associates
a signal with each state

• for each signal that she may receive, a belief about the states consis-
tent with the signal (a probability distribution over the set of states with
which the signal is associated)

• a Bernoulli payoff function over pairs (a,ω), where a is an action pro-

file and ω is a state, the expected value of which represents the player’s
preferences among lotteries over the set of such pairs.

Example: (An exchange game) Each of two individuals receives a

ticket on which there is an integer from 1 to m indicating the size of a
prize she may receive. The individuals’ tickets are assigned randomly and
independently; the probability of an individual’s receiving each possible
number is positive. Each individual is given the option to exchange her
prize for the other individual’s prize; the individuals are given this option
simultaneously. If both individuals wish to exchange then the prizes are
exchanged; otherwise each individual receives her own prize. Each indi-
13
vidual’s objective is to maximize her expected monetary payoff. Model
this situation as a Bayesian game and show that in any Nash equilibrium
the highest prize that either individual is willing to exchange is the small-
est possible prize.

EXAMPLE:A decision-maker in a single-person decision problem

cannot be worse off if she has more information: if she wishes, she can
ignore the information. In a game the same is not true: if a player
has more information and the other players know that she has more
information then she may be worse off. Consider, for example, the two-
player Bayesian game in Figure 4, where 0 < ϵ < 21 . In this game there
are two states, and neither player knows the state. Player 2’s unique
best response to every strategy of player 1 is L (which yields the expected
payoff 2-2(1-ϵ)p, whereas M and R both yield 32 − 32 (1−ϵ)p, where p is the
probability player 1 assigns to T), and player 1’s unique best response to
L is B. Thus (B, L) is the unique Nash equilibrium of the game, yielding
each player a payoff of 2

Now consider the variant of this game in which player 2 is informed of the
state: player 2’s signal function τ satisfies τ2(ω1) ̸= τ2(ω2). In this game
(T, (R, M)) is the unique Nash equilibrium. (Each type of player 2 has a
strictly dominant action, to which T is player 1’s unique best response.)
Player 2’s payoff in the unique Nash equilibrium of the original game is 2,
whereas her payoff in the unique Nash equilibrium of the game in which
she knows the state is 3ϵ in each state. Thus she is worse off when she

14
knows the state than when she does not. Player 2’s action R is good only
in state ω1 whereas her action M is good only in state ω2. When she
does not know the state she optimally chooses L, which is better than the
average of R and M whatever player 1 does. Her choice induces player 1 to
choose B. When player 2 is fully informed she optimally tailors her action
to the state, which induces player 1 to choose T. There is no steady state
in which she ignores her information and chooses L because this action
leads player 1 to choose B, making R better for player 2 in state ω1 and
M better in state ω2.

4 Cournot’s duopoly game with imperfect information:

Cournot’s duopoly model typically assumes perfect information, where

each firm knows the market demand curve and the cost structure of its
competitor. However, if you introduce imperfect information into the
model, it becomes more complex.
Imperfect information in Cournot’s duopoly game could mean that
the firms do not have complete knowledge about the cost structure or
demand curve of their competitors. This can lead to strategic uncertainty
and influence the firms’ decisions. Here are a few ways you might model
imperfect information in a Cournot’s duopoly.

15
4.1 Incomplete Information about Costs:

• Firms may not know the exact cost structure of their competitors.

• Each firm could have an estimate of the competitor’s costs, but these
estimates might be imprecise or outdated.

• Firms might need to make decisions under uncertainty, considering a

range of possible cost scenarios for their competitors.

4.2 Demand Uncertainty:

• Firms may have incomplete information about the market demand

curve.

• Each firm might have its own estimate of the demand curve, but
these estimates could differ.

• Firms might update their beliefs about the demand curve based on
observed market outcomes.

4.3 Strategic Information Gathering:

• Firms might invest in gathering information about their competitors,

incurring costs to reduce uncertainty.

• The cost of obtaining information could be a strategic variable that

firms decide on before determining their output levels.

16
4.4 Learning and Adaptation:

• Firms may learn from past interactions and adjust their strategies
over time.

• Learning could involve updating beliefs about competitors’ costs or

the demand curve based on observed market outcomes.

4.5 Information Asymmetry:

• One firm may have better information than the other, leading to an
information asymmetry.

• The firm with better information might exploit this advantage in its
production decisions.
Modeling Cournot’s duopoly with imperfect information often involves
game theory and the concept of Bayesian Nash equilibrium. In Bayesian
Nash equilibrium, players update their beliefs based on observed actions
and outcomes, and decisions are made in light of these beliefs.

The specifics of the model would depend on the nature of the imper-
fect information, the players’ strategies for dealing with uncertainty, and
how information is updated over time. The goal is to capture the strate-
gic interactions that arise when firms make decisions without complete
knowledge of their competitors’ costs or the market demand curve.

4.6 Example:

Let’s consider a simple example of Cournot’s duopoly with imperfect

information related to costs. In this scenario, each firm is uncertain about
the cost structure of its competitor. The firms produce identical goods
and simultaneously choose their output levels to maximize profits.
17
Assumptions:
• There are two firms, Firm A and Firm B.
• The market demand curve is Q = a − b(PA + PB ),Where Q is the
total quantity demanded,PAandPB are the prices changed by firms
A and B,and a and b are positive constants.

• Each firm has a constant marginal cost, but they are uncertain about
the exact value of the competitor’s marginal cost.

Let M CAandM CB be the marginal costs of firms A and B, respec-

tively. However, each firm believes that the competitor’s cost is drawn
from a distribution with mean µ and a certain level of uncertainty,
represented by the variance σ 2.

The profit function for firm A is:

πA − (PA − M CA).QA

Where QA is the output of Firm A. Similarly, the profit function for

Firm B is:

πA = (PB − M CB ).QB

The total quantity Q is determined by the sum of the individual

quantities produced by each firm:

Q=QA + QB

18
To model imperfect information,each firm has a belief about the com-
petitor’s cost,denoted as θA for firm A and θB for firm B.These beliefs
are drawn from the distribution with mean µ and variance σ 2.

The firms simultaneously choose their quantities to maximize ex-

pected profits given their beliefs about the competitor’s cost. The
equilibrium in this imperfect information scenario would involve find-
ing the Bayesian Nash equilibrium, where each firm’s strategy takes
into account its beliefs and the observed actions of the competitor.

This is a simplified example, and actual modeling would involve

more detailed assumptions and mathematical formulations based on
the specific nature of imperfect information in the duopoly setting.
Bayesian game theory provides a framework for analyzing these types
of situations where players have incomplete information.

19
 5 CONCLUSION

In conclusion, Bayesian games provide a powerful framework for mod-

eling strategic interactions under uncertainty. By incorporating Bayesian
reasoning into game theory, these games allow for a more realistic repre-
sentation of decision-making in situations where players have incomplete
information about each other’s characteristics or actions.
The key strength of Bayesian games lies in their ability to capture the
dynamic nature of information in strategic interactions. Players update
their beliefs based on observed actions and outcomes, leading to a richer
understanding of how information influences strategic choices. This mod-
eling approach is particularly relevant in real-world scenarios where agents
must make decisions without complete knowledge of the environment or
the intentions of other players.
The application of Bayesian games extends across various fields, includ-
ing economics, political science, and artificial intelligence. The framework
facilitates a nuanced analysis of strategic behavior in situations where
uncertainty is a fundamental aspect of decision-making.However, it’s im-
portant to acknowledge the challenges associated with Bayesian games,
such as computational complexity and the need for sophisticated algo-
rithms to solve them. Despite these challenges, the insights gained from
Bayesian game models contribute significantly to our understanding of
strategic interactions in complex and uncertain environments.
In summary, Bayesian games offer a valuable tool for studying strategic
decision-making in situations where uncertainty plays a crucial role. The
continued development of this framework promises to enhance our abil-
ity to analyze and predict strategic interactions in diverse and dynamic
settings.

20
6 REFERENCE:

• ”Game Theory: An Introduction” by Eilon Solan, Nimrod Megiddo,

and Adam Shwartz.

• ”Game Theory and the Social Contract: Just Playing” by Ken Bin-
more.

• ”A Course in Game Theory” by Martin J. Osborne and Ariel Rubin-

stein.

• ”Game Theory: Analysis of Conflict” by Roger B. Myerson.

• ”Multiagent Systems: Algorithmic, Game-Theoretic, and Logical

Foundations” by Yoav Shoham and Kevin Leyton-Brown.

• ”Information and Learning in Markets: The Impact of Market Mi-

crostructure” by Xavier Vives.

• ”Bayesian Games” by Guofu Tan and Susheng Wang.

• ”Algorithmic Game Theory” by Noam Nisan, Tim Roughgarden, Eva

Tardos, and Vijay V. Vazirani.

• ”An Introduction to Game Theory” by Martin J.Osborne.

• ”Evolutionary Game Theory” by Josef Hofbauer and Karl Sigmund.

21