Game Theory

by
Purbita Jana
ii
Contents

Introduction v

1 History of Game Theory 1

2 Normal form Game 3

3 Appendix 19
iv CONTENTS
 Introduction

This article is
vi CONTENTS
Chapter 1

History of Game Theory

2 History of Game Theory
Chapter 2

Normal form Game

Definition 2.0.1 (Strategic game/ Normal form game). A strategic game or

normal form game is a triple (N, {Ai }i∈N , {ui }i∈N ) consisting of

• a finite set, N , of players or agents.

• for each player i ∈ N , a nonempty set Ai representing set of actions or

strategies available to player i.

• player i’s payoff is a function ui : A1 × A2 × · · · × An (:= A) → R,

where ui (a) is i’s payoff if a(:= (a1 , a2 , · · · , an )) is the profile of actions
or strategies.

Definition 2.0.2 (Action Profile). Let (N, {Ai }i∈N , {ui }i∈N ) be a normal form
game. Action profile/ profile of actions a = (a1 , a2 , · · · , an ) is an ordered tuple
where a ∈ A1 × A2 × · · · × An (:= A).

Example 2.0.1. The payoff matrix is given by the following matrix:

4 Normal form Game

Player 2

H T
H (1, −1) (−1, 1)
Player 1
T (−1, 1) (1, −1)
Here N = {1, 2}, A1 = {H, T }, A2 = {H, T },

A = A1 × A2

= {H, T } × {H, T }

= {(H, H), (H, T ), (T, H), (T, T )}.

The utility function for both the players for this game is given as,

u1 :A → R u2 :A → R

A1 × A2 → R A1 × A2 → R

(H, H) 7→ 1 (H, H) 7→ −1

(H, T ) 7→ −1 (H, T ) 7→ 1

(T, H) 7→ −1 (T, H) 7→ 1

(T, T ) 7→ 1 (T, T ) 7→ −1.

Example 2.0.2. Consider the game with the following pay off matrix:

Player 2

C D
.
C (−5, −5) (0, −10)
Player 1
D (−10, 0) (−1, −1)
Here N = {1, 2}, A1 = {C, D}, A2 = {C, D},

A = A1 × A2

= {C, D} × {C, D}

= {(C, C), (C, D), (D, C), (D, D)}.

5 Normal form Game

The utility function for both the players for this game is given as,

u1 :A → R u2 :A → R

A1 × A2 → R A1 × A2 → R

(C, C) 7→ −5 (C, C) 7→ −5

(C, D) 7→ 0 (C, D) 7→ −10

(D, C) 7→ −10 (D, C) 7→ 0

(D, D) 7→ −1 (D, D) 7→ −1

Dominant Strategy

A strategy that always achieves a better result for a given player, no matter
what the other player does, is a dominant strategy. In other words, a dominant
strategy for a player is one that produces the highest payoff of any strategy
available for every possible actions by other players.

Definition 2.0.3 (Dominant Strategy). Let (N, {Ai }i∈N , {ui }i ) be a normal
form game. A strategy ai ∈ Ai is dominant strategy for player i if

ui (ai , a−i ) ≥ ui (a′i , a−i )

for all a−i ∈ A−i , and a′i ∈ Ai , where A−i = A1 ×A2 ×· · ·×Ai−1 ×Ai+1 ×· · ·×An
and (ai , a−i ) := (a1 , a2 , · · · , an ) = a and (a′i , a−i ) := (a1 , a2 , · · · , ai−1 , a′i , ai+1 , · · · , an ).

Definition 2.0.4 (Strictly Dominant Strategy). Let (N, {Ai }i∈N , {ui }i ) be a
normal form game. A strategy ai ∈ Ai is strictly dominant strategy for player
i if
ui (ai , a−i ) > ui (a′i , a−i )

for all a−i ∈ A−i , and a′i (̸= ai ) ∈ Ai , where A−i = A1 × A2 × · · · × Ai−1 ×
Ai+1 × · · · × An .
6 Normal form Game

Example 2.0.3. The payoff matrix is given by the following matrix:

Player 2

C D
C (−5, −5) (0, −10)
Player 1
D (−10, 0) (−1, −1)

Here N = {1, 2}, A1 = {C, D}, A2 = {C, D},

A = A1 × A2

= {C, D} × {C, D}

= {(C, C), (C, D), (D, C), (D, D)},

u1 :A → R u2 :A → R

A1 × A2 → R A1 × A2 → R

(C, C) 7→ −5 (C, C) 7→ −5

(C, D) 7→ 0 (C, D) 7→ −10

(D, C) 7→ −10 (D, C) 7→ 0

(D, D) 7→ −1 (D, D) 7→ −1.

Here for Player 1, u1 (C, C) > u1 (D, C) and u1 (C, D) > u1 (D, D). Hence C is
the strictly dominant strategy for Player 1. Similarly one may conclude that C
is strictly dominant strategy for Player 2.

Exercise 1. Find the dominant and strictly dominant strategy (if any) for each
players for the following games described by the following payoff matrices.

1. Matching Pennies:
7 Normal form Game

Player 2

H T
H (1, −1) (−1, 1)
Player 1
T (−1, 1) (1, −1)

2. Prisoner’s Dilemma:

Player 2

C D
C (−5, −5) (0, −10)
Player 1
D (−10, 0) (−1, −1)

3. Predator Prey:

Player 2

D H
D (3, 3) (1, 4)
Player 1
H (4, 1) (0, 0)

4. Stag Hare:

Player 2

S H
S (5, 5) (0, 3)
Player 1
H (3, 0) (4, 4)

5. Battle of Sexes:

Player 2

X Y
X (3, 1) (0, 0)
Player 1
Y (0, 0) (1, 3)
8 Normal form Game

6. Run Stop 1:

Player 2

R S
R (3, 4) (0, 5)
Player 1
S (4, 0) (1, 1)

7. Run Stop 2:

Player 2

R S
R (1, 1) (−1, −3)
Player 1
S (2, −1) (2, 3)

8. Run Stop 3:

Player 2

R S
R (−5, −5) (1, 0)
Player 1
S (0, 1) (−1, −1)

Definition 2.0.5 (Best Reply). Let (N, {Ai }i∈N , {ui }i∈N ) be a normal form
game. A strategy ai is a best reply, also known as best response of player i to
a profile of strategies a−i ∈ A−i for the other players if

ui (ai , a−i ) ≥ ui (a′i , a−i )

for all a′i ∈ Ai .

Definition 2.0.6 (Nash Equilibrium). A Nash equilibrium of a strategic game

(N, {Ai }i∈N , {ui }i∈N ) is a profile a∗ ∈
Q
Ai (:= A1 × A2 × · · · × An ) of actions
i∈N
with the property that for all players i ∈ N we have

ui (a∗−i , a∗i ) ≥ ui (a∗−i , ai ), for all ai ∈ Ai .

9 Normal form Game

Note: A profile of dominant strategies is a Nash equilibrium but not vice

versa.

Exercise 2. Find the pure strategy Nash equilibrium for the following games
described by the following payoff matrices.

1. Matching Pennies:

Player 2

H T
H (1, −1) (−1, 1)
Player 1
T (−1, 1) (1, −1)

2. Prisoner’s Dilemma:

Player 2

C D
C (−5, −5) (0, −10)
Player 1
D (−10, 0) (−1, −1)

3. Predator Prey:

Player 2

D H
D (3, 3) (1, 4)
Player 1
H (4, 1) (0, 0)

4. Stag Hare:

Player 2

S H
S (5, 5) (0, 3)
Player 1
H (3, 0) (4, 4)
10 Normal form Game

5. Battle of Sexes:

Player 2

X Y
X (3, 1) (0, 0)
Player 1
Y (0, 0) (1, 3)

6. Run Stop 1:

Player 2

R S
R (3, 4) (0, 5)
Player 1
S (4, 0) (1, 1)

7. Run Stop 2:

Player 2

R S
R (1, 1) (−1, −3)
Player 1
S (2, −1) (2, 3)

8. Run Stop 3:

Player 2

R S
R (−5, −5) (1, 0)
Player 1
S (0, 1) (−1, −1)

Proposition 2.0.1. Pure strategy Nash equilibrium for a finite game may or
may not exist and furthermore when it exists, it may or may not be unique.
11 Normal form Game

It is to be noted that John Nash received the Nobel Prize mainly to show
that any finite game with (a finite number of players and a finite number of
strategies) has an equilibrium point. That equilibrium may not exist in pure
strategy but existence is guaranteed while allowing mixed strategies. That is
allowing probability distributions over strategies.

Definition 2.0.7 (Mixed Strategy). Let (N, {Ai }i∈N , {ui }i∈N ) be a strategic
game. A mixed strategy/action of player i ∈ N is a probability distribution
over his/her strategies / actions Ai and denoted by
 
 X 
Si = si : Ai → [0, 1] si (aj ) = 1 .
 
aj ∈Ai

Example 2.0.4. Let us consider the matching pennies game with probability
distribution {(p, 1 − p) | p ∈ [0, 1]} for Player 1’s startegies and {(q, 1 − q) |
q ∈ [0, 1]} for Player 2’s strategies.

Player 2

q 1−q
H T
p H (1, −1) (−1, 1)
Player 1
1−p T (−1, 1) (1, −1)

Here A1 = {H, T }, A2 = {H, T }, A1 × A2 = {(H, H), (H, T ), (T, H), (T, T )},
s1 (H) = p, s2 (H) = q, s1 (T ) = 1 − p, s2 (T ) = 1 − q, where p ∈ [0, 1] and
q ∈ [0, 1].
 
 X 
S1 = s1 : A1 → [0, 1] s1 (aj ) = 1
 
aj ∈A1

= {(p, 1 − p) | p ∈ [0, 1]}.

12 Normal form Game

 
 X 
S2 = s2 : A2 → [0, 1] s2 (aj ) = 1
 
aj ∈A2

= {(q, 1 − q) | q ∈ [0, 1]}.

Definition 2.0.8 (Mixed Strategy Extension). Let G = (N, {Ai }i∈N , {ui }i∈N )
be a normal form game. The mixed extension of G is the game

Γ = (N, {Si }i∈N , {Ui }i∈N ),

Q
where Si is the mixed strategy for player i and Ui : Si → R such that
i∈N
X
Ui (s) = ui (a1 , a2 , · · · , an )s1 (a1 )s2 (a2 ), · · · sn (an )
(a1 ,a2 ,··· ,an )∈A
!!
X Y
i.e, Ui (s) = ui (a) sj (aj ) .
a∈A j∈N

Example 2.0.5. Let us consider the matching pennies game with probability
distribution {(p, 1 − p) | p ∈ [0, 1]} for Player 1’s startegies and {(q, 1 − q) |
q ∈ [0, 1]} for Player 2’s strategies.

Player 2

q 1−q
H T
p H (1, −1) (−1, 1)
Player 1
1−p T (−1, 1) (1, −1)
Here A1 = {H, T }, A2 = {H, T }, A1 × A2 = {(H, H), (H, T ), (T, H), (T, T )},
s1 (H) = p, s2 (H) = q, s1 (T ) = 1 − p, s2 (T ) = 1 − q, where p ∈ [0, 1] and
q ∈ [0, 1].
 
 X 
S1 = s1 : A1 → [0, 1] s1 (aj ) = 1
 
aj ∈A1

= {(p, 1 − p) | p ∈ [0, 1]}.

13 Normal form Game

 
 X 
S2 = s2 : A2 → [0, 1] s2 (aj ) = 1
 
aj ∈A2

= {(q, 1 − q) | q ∈ [0, 1]}.

U1 : S1 × S2 → R
X
((p, 1 − p), (q, 1 − q)) 7→ u1 (a)s1 (a1 )s2 (a2 )
a∈A

= u1 (H, H)s1 (H)s2 (H) + u1 (H, T )s1 (H)s2 (T )

+ u1 (T, H)s1 (T )s1 (H) + u1 (T, T )s1 (T )s2 (T )

= 4pq − 2p − 2q + 1

U2 : S1 × S2 → R
X
((p, 1 − p), (q, 1 − q)) 7→ u2 (a)s1 (a1 )s2 (a2 )
a∈A

= u2 (H, H)s1 (H)s2 (H) + u2 (H, T )s1 (H)s2 (T )

+ u2 (T, H)s1 (T )s1 (H) + u2 (T, T )s1 (T )s2 (T )

= −4pq + 2p + 2q − 1.

It is to be noted that for our convenience we may denote Ui ((p, 1 − p), (q, 1 − q))
as Ui (p, q).

Definition 2.0.9 (Mixed strategy Nash equilibrium). A profile of mixed strate-

gies s∗ = (s∗1 , s∗2 , · · · , s∗n ) forms a mixed strategy Nash equilibrium if

Ui (s∗ ) ≥ Ui (s∗−i , si ), for all i and si ∈ Si .

! !
X Y X Y
i.e., ui (a∗ ) sj (a∗j ) ≥ ui (a∗−i , ai ) sj (aj ) ,
a∗ ∈A j∈N a−i ∈A−i j̸=i

for all i and si ∈ Si .

Example 2.0.6. Let us consider the matching pennies game with probability
distribution {(p, 1 − p) | p ∈ [0, 1]} for Player 1’s startegies and {(q, 1 − q) |
q ∈ [0, 1]} for Player 2’s strategies.
14 Normal form Game

Player 2

q 1−q
H T
p H (1, −1) (−1, 1)
Player 1
1−p T (−1, 1) (1, −1)

Here A1 = {H, T }, A2 = {H, T }, A1 × A2 = {(H, H), (H, T ), (T, H), (T, T )},
s1 (H) = p, s2 (H) = q, s1 (T ) = 1 − p, s2 (T ) = 1 − q, where p ∈ [0, 1] and
q ∈ [0, 1].

 
 X 
S1 = s1 : A1 → [0, 1] s1 (aj ) = 1
 
aj ∈A1

= {(p, 1 − p) | p ∈ [0, 1]}.

 
 X 
S2 = s2 : A2 → [0, 1] s2 (aj ) = 1
 
aj ∈A2

= {(q, 1 − q) | q ∈ [0, 1]}.

15 Normal form Game

U1 : S1 × S2 → R
X
((p, 1 − p), (q, 1 − q)) 7→ u1 (a)s1 (a1 )s2 (a2 )
a∈A

= u1 (H, H)s1 (H)s2 (H) + u1 (H, T )s1 (H)s2 (T )

+ u1 (T, H)s1 (T )s1 (H) + u1 (T, T )s1 (T )s2 (T )

= 4pq − 2p − 2q + 1

U2 : S1 × S2 → R
X
((p, 1 − p), (q, 1 − q)) 7→ u2 (a)s1 (a1 )s2 (a2 )
a∈A

= u2 (H, H)s1 (H)s2 (H) + u2 (H, T )s1 (H)s2 (T )

+ u2 (T, H)s1 (T )s1 (H) + u2 (T, T )s1 (T )s2 (T )

= −4pq + 2p + 2q − 1.

It is to be noted that for our convenience we may denote Ui ((p, 1 − p), (q, 1 − q))
as Ui (p, q). Let (p∗ , q ∗ ) be mixed strategy Nash equilibrium. Then,

U1 (p∗ , q ∗ ) ≥ U1 (p, q ∗ ), for all p ∈ [0, 1]

and

U2 (p∗ , q ∗ ) ≥ U1 (p∗ , q), for all q ∈ [0, 1]

Hence we have,

U1 (p∗ , q ∗ ) ≥ U1 (p, q ∗ ), for all p ∈ [0, 1]

4p∗ q ∗ − 2p∗ − 2q ∗ + 1 ≥ 4pq ∗ − 2p − 2q ∗ + 1, for all p ∈ [0, 1]

2q ∗ (p∗ − p) − (p∗ − p) ≥ 0, for all p ∈ [0, 1]

(2q ∗ − 1)(p∗ − p) ≥ 0, for all p ∈ [0, 1] −→ (1)

16 Normal form Game

Similarly,

U2 (p∗ , q ∗ ) ≥ U2 (p∗ , q), for all q ∈ [0, 1]

−4p∗ q ∗ + 2p∗ + 2q ∗ − 1 ≥ −4p∗ q + 2p∗ + 2q − 1, for all q ∈ [0, 1]

2p∗ (q − q ∗ ) − (q − q ∗ ) ≥ 0, for all q ∈ [0, 1]

(2p∗ − 1)(q − q ∗ ) ≥ 0, for all q ∈ [0, 1] −→ (2)

Now to calculate the values of p∗ and q ∗ from the inequalities (1) and (2) we
may consider the following three conditions:

1. p∗ = 0

For this case we get from (1), −p(2q ∗ − 1) ≥ 0, for all p ∈ [0, 1]. Hence
2q ∗ − 1 ≤ 0 and so q ∗ ≤ 21 .

From 2 we have, q ∗ − q ≥ 0, for all q ∈ [0, 1]. Hence q ∗ ≥ q, for all

q ∈ [0, 1] gives q ∗ = 1. But it contradicts the fact that q ∗ ≤ 21 . Hence p∗
can’t be 0

2. p∗ = 1

For this case we get from (1), (2q ∗ − 1)(1 − p) ≥ 0, for all p ∈ [0, 1].
Hence 2q ∗ − 1 ≥ 0 and so q ∗ ≥ 12 .

From 2 we have, q −q ∗ ≥ 0, for all q ∈ [0, 1]. Hence q ≥ q ∗ , for all q ∈ [0, 1]
gives q ∗ = 0. But it contradicts the fact that q ∗ ≥ 12 . Hence p∗ can’t be 1

3. p∗ ∈ (0, 1)

Then, for a fixed p∗ ∈ (0, 1), there exist p1 ∈ [0, 1] such that (p∗ − p1 ) ≥ 0
and p2 ∈ [0, 1] such that (p∗ − p2 ) ≤ 0. Hence the only possibility to
satisfy (1) for all p ∈ [0, 1] is 2q ∗ − 1 = 0. Hence q ∗ = 12 .

Similarly from (2) we have p∗ = 12 .

17 Normal form Game

Hence the mixed strategy equilibrium for this game is given by (( 12 , 12 ), ( 12 , 21 )).

Exercise 3. Find the mixed strategy Nash equilibrium for the following games
described by the following payoff matrices.

1. Prisoner’s Dilemma:

Player 2

C D
C (−5, −5) (0, −10)
Player 1
D (−10, 0) (−1, −1)

2. Predator Prey:

Player 2

D H
D (3, 3) (1, 4)
Player 1
H (4, 1) (0, 0)

3. Stag Hare:

Player 2

S H
S (5, 5) (0, 3)
Player 1
H (3, 0) (4, 4)

4. Battle of Sexes:

Player 2

X Y
X (3, 1) (0, 0)
Player 1
Y (0, 0) (1, 3)
18 Normal form Game

5. Run Stop 1:

Player 2

R S
R (3, 4) (0, 5)
Player 1
S (4, 0) (1, 1)

6. Run Stop 2:

Player 2

R S
R (1, 1) (−1, −3)
Player 1
S (2, −1) (2, 3)

7. Run Stop 3:

Player 2

R S
R (−5, −5) (1, 0)
Player 1
S (0, 1) (−1, −1)
Chapter 3

Appendix

Definition 3.0.1 (Set). A set is a collection of well defined objects.

Definition 3.0.2 (Cartesian Product). Let A and B be two non-empty sets.

The Cartesian product of A and B is the set of all ordered pairs and denoted
by A × B, i.e.,
A × B = {(a, b) | a ∈ A and b ∈ B}.

Definition 3.0.3 (Relation). Let A and B be two sets. A relation say R

between A and B is a subset of A × B.

Definition 3.0.4 (Function). Let A and B be two non-empty sets. A function

f from A to B is a relation that assigns each element x of A, an unique element
y in B. A is said to be the domain of f and B is said to be the codomain of f
and denoted by f : A → B.
The set {f (x) | x ∈ A} is said to be the range of f and denoted by f (A)
whereas {x | f (x) = y} is the preimage of y ∈ B under f : A → B and is
denoted by f −1 (y).

Definition 3.0.5 (Sample Space). The set of all possible outcomes of an ex-
periment is known as the sample space of the experiment and is denoted by
20 Appendix

S.

For example the experiment of tossing coin consists of the sample space
S = {H, T }, where H stands for head and T stands for tail.

Definition 3.0.6 (Event). Any subset E of the sample space S is known as

an event.

For example {T } is an event for tossing coin example. As the sample space
for the experiment was {H, T } and {T } ⊆ {H, T }. Similarly {H} is also an
event.

Definition 3.0.7 (Probability). Consider an experiment whose sample space

is S. For each event E of the sample space S, we assume that a number P (E)
is defined and satisfies the following three conditions:

1. 0 ≤ P (E) ≤ 1

2. P (S) = 1

3. For any sequence of events E1 , E2 , . . . that are mutually exclusive, that

is, events for which En ∩ Em = ∅ when n ̸= m, then
∞
[ ∞
X
P( En ) = P (En ).
n=1 n=1

We refer to P (E) as the probability of the event E.

In the coin tossing example, if we assume that a head is equally likely to

appear as a tail, then we would have P ({H}) = P ({T }) = 12 .