Professional Documents
Culture Documents
Sanjay Singh
1
Department of Information and Communication Technology
Manipal Institute of Technology, MAHE
Manipal-576104, INDIA
sanjay.singh@manipal.edu
2
Centre for Artificial and Machine Intelligence (CAMI)
MAHE, Manipal-576104, INDIA
Module Objective
Nash equilibrium
Maxmin value
Minmax value
Maxmin strategy
Minmax strategy
Subgame perfect equilibrium
Nash Equilibrium
Alternatively
2
1
A B
A 2,1 0,0
B 0,0 1,2
2
1
NC C
NC -2,-2 -10,-1
C -1,-10 -5,-5
Tragedy of Commons
n−5 5X
ui (s1 , . . . , sn ) = si − sj (i = 1, 2, . . . , n)
n n
j6=i
When n < 5, (0, 0, . . . , 0) is a strongly dominant strategy
equilibrium
When n > 5, (1, 1, . . . , 1) is a strongly dominant strategy
equilibrium
5X
When n = 5, ui (0, s−i ) = ui (1, s−i ) = − si
n
j6=i
Thus, ui (0, s−i ) = ui (1, s−i ), ∀s−i ∈ S−i ∀i ∈ N
It implies that bi (s−i ) = {0, 1} ∀s−i ∈ S−i ∀i ∈ N
Here all the strategy profiles are Nash equilibrium
5X 5 5X
ui (0, s−i ) = − sj , ui (1, s−i ) = −4 − − sj
n n n
j6=i j6=i
It implies that bi (s−i ) = {0} ∀i ∈ N
Hence, whatever be the value of n, profile (0, 0, . . . , 0) is a
strongly dominant strategy equilibrium
n
The total payoff to all players combined is
(1 + n)2
1 1
Now lets consider a non-equilibrium profile , . . . ,
2n 2n
With such a profile each player gets a payoff
1 n 1
1− =
2n 2n 4n
1 n
The total payoff for all the players = >
4 (1 + n)2
Thus a non-equilibrium profile provides higher payoff than
a Nash equilibrium payoff
In general, like in the PDP, the equilibrium payoffs may not
be the best possible outcome for the players individually or
collectively or both
It is a limitation that Nash equilibrium payoff suffers often
A B C D Rmin
A 4 3 2 5 2 ← maxmin
B -10 2 0 -1 -10
C 7 5 2 3 2 ← maxmin
D 0 8 -4 -5 -5
Cmax 7 8 2 ↑ minmax 5
Proposition
Suppose a strategic form game Γ = hN, (Si ), (ui )i has a pure
strategy Nash equilibrium (s∗1 , . . . , s∗n ), then
ui (s∗1 , . . . , s∗n ) ≥ vi ∀i ∈ N
Proof.
We note that ∀i ∈ N,
Any strategy profile s∗−i ∈ S−i of the players that forces the
payoff vi on player i is called a minmax strategy profile (of the
rest of the player) against player i.
2
1
A B
A 4,1 0,4
B 1,5 1,1
Proposition
Consider a strategic form game Γ = hN, (Si ), (ui )i. Then
vi ≥ vi ∀i ∈ N.
Proof.
Suppose s∗−i is a minmax strategy against player i. This means
Note ∀i ∈ N that
Thus the minmax value of a player is no less than his maxmin value.
ui (s∗1 , . . . , s∗n ) ≥ vi ∀i ∈ N.
Proof.
First, we note that ∀i ∈ N,
The payoff of a player in a pure strategy Nash equilibrium (if one exists) is
greater than or equal to the minmax value of the player which in turn is
greater than or equal to the maxmin value of the player
2
1
E F
CG 1,2 3,1
CH 0,0 3,1
DG 2,0 2,0
DH 2,0 2,0
2
1
accept fight
in 2,1 0,0
out 1,2 1,2