41

Pareto Ranked Coordination
G
B
G
2,2
0,0
B
0,0
1,1
3 equilibria:
(
B,B
)
,
(
O,O
)
, and
((
1 3
,
2 3
)
,
(
1 3
,
2 3
))
42

A Special Game
G
B
G
1,0
-1,0
B
0,0
0,0
How many equilibria?
43

A Special Game
G
B
G
1,0
-1,0
B
0,0
0,0
Infinite equilibria:
((
a,
1
-
a
)
,
(
1 2
,
1 2
))
for any
a
∈
[
0
,
1
]
44

45

46


48

49

Fact 2
For any game, the set of mixed strategy profiles is convex. For a
finite game, the set of mixed strategy profiles is compact.
Fact 3
The best response correspondence is convex valued. For a finite
game, the best response correspondence is upper hemi-continuous and
compact valued.
Thus, in a nutshell, allowing for mixed strategies guarantees the existence
of Nash equilibria in finite games.
50

Minmax and Interchangeability
Theorem 4 (Minmax Theorem)
For every two-person, zero-sum game with
finite pure strategies, there exists a value
V
and a mixed strategy for each
player, such that (a) Given player 2’s strategy, the best payoff possible for
player 1 is
V
, and (b) Given player 1’s strategy, the best payoff possible for
player 2 is
-
V
.
Interchangeability.
All equilibrium strategies for either given player in a
two-player constant sum game are equivalent and any pair of equilibrium
strategies is an equilibrium:
For a two player, constant-sum game, if
(
s
*
1
,s
*
2
)
is an equilibrium and
(
s
**
1
,s
**
2
)
is an equilibrium, then
(
s
*
1
,s
**
2
)
and
(
s
**
1
,s
*
2
)
are also equilib-
ria.
51

How Many Equilibria?
The Oddness Theorem
Theorem 5 (Wilson (1971))
Almost all finite games have a finite and odd
number of equilibria.
(“Almost all” means it is not the case for a set of games possessing Lebesqgue
measure zero in
R
∏
i
∈
N
Ai
.)
52

The Generic Oddness of Fixed Points
53

The Generic Oddness of Fixed Points
54

Equilibrium Refinements
●
Pareto dominance/efficiency
●
Symmetry
Consider Battle of the Sexes: symmetric equilibrium is Pareto dominated.
55

Iterated Dominance
p
-Beauty Contest
Suppose that 3 individuals each simultaneously an-
nounce a number
x
i
∈
[
0
,
100
]
. Player
i
’s payoff is
u
i
(
a
1
,a
2
,a
3
)
=
1
if
a
i
-
p
3
∑
3
i
=
1
a
i
<
min
j
≠
i
a
j
-
p
3
∑
3
i
=
1
a
i
0
if
a
i
-
p
3
∑
3
i
=
1
a
i
>
min
j
≠
i
a
j
-
p
3
∑
3
i
=
1
a
i
(and the prize is split evenly if more than one individual wins.)
56

You've reached the end of your free preview.
Want to read all 57 pages?
- Spring '11
- JohnPaddy
- Game Theory, player, Nash, best response, pure strategies