Game Theory
-
Upload
ayushman-shukla -
Category
Documents
-
view
214 -
download
0
Transcript of Game Theory
![Page 1: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/1.jpg)
DYNAMIC GAMES OF COMPLETE INFORMATION
PRELIMINARIES
![Page 2: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/2.jpg)
The Extensive-‐Form Game • Set of players N • Players’ payoffs as a function of outcomes 𝑣!(. ) !∈! • Order of moves • Actions of players when they can move • The knowledge that players have when they can move. • Probability distributions against exogenous events. • The structure of the extensive-‐form game represented by the above mentioned points is common knowledge among all players.
![Page 3: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/3.jpg)
Game Trees
Definition: A game tree is a set of nodes 𝑥 ∈ 𝑋 with a precedence relation 𝑥 > 𝑥′, which means “x precedes x’”. Every node in a game tree has only one predecessor. The precedence relation is transitive, asymmetric and incomplete. There is a special node called the root of the tree, denoted by x0, which precedes any other 𝑥 ∈ 𝑋. Nodes that do not precede any other nodes are called terminal nodes, denoted by the set 𝑍 ⊂ 𝑋. Terminal nodes denote the final outcomes of the game with which payoffs are associated. Every node x that is not a terminal node is assigned to either a player or to Nature.
1
x0
o f o f
F O
x1 2
x2 2
2 1
0 0
0 0
1 2
x3 x4 x5
x6
![Page 4: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/4.jpg)
Information Sets
Definition: Every player i has a collection of information sets ℎ! ∈ 𝐻! that partition the nodes of the game at which player i moves with the following properties: 1. If ℎ! is a singleton that includes only x then player i who moves at x knows that he is at x.
2. If 𝑥 ≠ 𝑥′ and if both 𝑥 ∈ ℎ! and 𝑥! ∈ ℎ! then player i who moves at x does not know whether he is at x or x’.
3. If 𝑥 ≠ 𝑥′ and if both 𝑥 ∈ ℎ! and 𝑥! ∈ ℎ! then 𝐴! 𝑥 = 𝐴!(𝑥′).
o f o f
F O
x1 2
x2 2
2 1
0 0
0 0
1 2
x3 x4 x5 x6
1
x0
![Page 5: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/5.jpg)
Imperfect versus Perfect Information Definition: A game of complete information in which every information set is a singleton and there are no moves of Nature is called a game of perfect information. A game in which some information sets contain several nodes or in which there are moves of Nature is called a game of imperfect information. Imperfect information can be due to two causes: • Exogenous uncertainty: Uncertainty over the choice of Nature.
• Endogenous uncertainty: Uncertainty over the choice of another player.
![Page 6: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/6.jpg)
Strategies and Nash Equilibrium Pure Strategies in Extensive-‐Form Games: A pure strategy for player i is a complete plan of play that describes which pure action player i will choose at each of his information sets. In the simultaneous move Battle of Sexes game Si={O,F} for both players. In the sequential move Battle of Sexes game, • S1={O,F} • S2={oo, of, fo, ff}
Definition: A pure strategy for player i is a mapping 𝑠!: 𝐻! → 𝐴! that assigns an action 𝑠!(ℎ!) ∈ 𝐴!(ℎ!) for every information set ℎ! ∈ 𝐻! . We denote by 𝑆! the set of all pure strategy mappings 𝑠! ∈ 𝑆!
![Page 7: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/7.jpg)
Mixed versus Behavioral Strategies Definition: A mixed strategy for player i is a probability distribution over his pure strategies 𝑠! ∈ 𝑆! Definition: A behavioral strategy specifies for each information set ℎ! ∈ 𝐻! an independent probability distribution over 𝐴!(ℎ!) and is denoted by 𝜎!: 𝐻! → ∆𝐴!(ℎ!) where 𝜎!(𝑎!(ℎ!)) is the probability that player i plays action 𝑎!(ℎ!) ∈ 𝐴!(ℎ!) in information set ℎ! .
Pr{o|O} = poo + pof = 1/3 ; 𝜎! 𝑜 ℎ!! = !
!
Pr{f|O} = pfo + pff = 2/3 ; 𝜎! 𝑓 ℎ!! = !
!
Pr{o|F} = poo + pfo = 1/2 ; 𝜎! 𝑜 ℎ!! = !
!
Pr{f|F} = pof + pff = 1/2 ; 𝜎! 𝑓 ℎ!! = !
!
poo + pfo + pof + pff = 1
o f o f
F O
x1 2
x2 2
2 1
0 0
0 0
1 2
1/3 2/3 1/2 1/2
x0
1
![Page 8: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/8.jpg)
Definition: A game of perfect recall is one in which no player ever forgets information that he previously knew. The Absent Minded Driver Introduced by Piccione and Rubinstein in 1997 Single person decision problem. A driver has to go home. Exit 1 is the wrong exit and gives him a payoff 0 Exit 2 is the correct exit and gives him a payoff 4. Missing both exits will eventually lead him home through a long way, giving him a payoff of 1. Mixed strategy: The probability that he will exit at any Exit he passes in p. Expected payoff: 0𝑝 + 4𝑝 1 − 𝑝 + 1(1 − 𝑝)! Maximized for 𝑝 = !
!
If he knows with some probability q that he s at Exit 1 and with some probability (1-‐q) that he is at Exit 2, his expected payoff would be 𝑞 4𝑝 1 − 𝑝 + 1 1 − 𝑝 ! + 1 − 𝑞 [4𝑝 +1(1 − 𝑝)] This leads to an interesting dynamic inconsistency: The driver would plan one thing in advance and then rationally change his mind once he finds himself at an intersection.
1
1 Exit
Exit
1 4
0
![Page 9: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/9.jpg)
Normal-‐Form Representation of Extensive-‐Form Games Simultaneous move Battle of Sexes game Chris Alex
O
F
O
2,1
0,0
F
0,0
1,2
Sequential move Battle of Sexes game Chris Alex
oo
of
fo
ff
O
2,1
2,1
0,0
0,0
F
0,0
1,2
0,0
1,2
Even though we have been able to represent the extensive form game into the normal form game, there is the lack of dynamism in the representation. Why do we then try to represent it in this way? Concept of Nash equilibrium is static in nature, in that the equilibrium posits that players take the strategies of others as given, and in turn they play a best response. The above representation suffices to find all Nash equilibria of the game.
![Page 10: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/10.jpg)
Nash Equilibrium and Paths of Play Consider the normal form representation of the sequential Battle of Sexes game. Chris Alex
oo
of
fo
ff
O
2,1
2,1
0,0
0,0
F
0,0
1,2
0,0
2,1
Nash equilibria: (O,oo), (O,of), (F,ff) Even though (O,oo) and (O,of) result in the exact same outcome, there is an important difference. The difference between these two equilibria is not in what the players actually play in equilibrium, but instead what player 2 plans to play in an information set that is not reached in equilibrium.
![Page 11: Game Theory](https://reader034.fdocuments.us/reader034/viewer/2022051708/577ccd521a28ab9e788bfdf0/html5/thumbnails/11.jpg)
In the extensive form of the game, every outcome is associated with a unique path of play. If a strategy profile s* is a Nash equilibrium, each player i prefers sticking to the predicted path of play over leaving it and choosing some other path in the game, given his belief that the other players will stick to 𝑠!!∗ Let 𝜎∗ = (𝜎!∗,𝜎!∗,… 𝜎!∗) be a Nash equilibrium profile of behavioral strategies in an extensive-‐form game. We say that an information set is on the equilibrium path if given 𝜎∗it is reached with positive probability, We say that an information set is off the equilibrium path if given 𝜎∗ it is never reached
o f o f
F O
x1 2
x2 2
2 1
0 0
1 2
x3 x4 x5 x6
x0
1