الفهرس 
Discrete-Time Markov ChainsOnly 14 pages are availabe for public view
 |  | from | 150 |  |  |
| | | from | 150 | | |
Abstract
Although, most game theory researches on the prisoner’s
dilemma have centered on two-player models, it
is possible to create it to be consisted of three or even
more players. In this thesis, I am interested in the model
of three-player iterated prisoner’s dilemma game where,
each player has two choices. The action of each strategy
in this model depends on the previous action of the last
round. Each strategy is presented by finite state of automata.
I used a computer program to calculate the payoff
values resulting from the actions of all possible strategies.
I studied the behavior of four different strategies
related to Tit for Tat concept. The conditions of each strategy
to be the best are determined. Due to the computational
advantage in symmetric games, most research has
focused on the symmetric games instead of the asymmetric
ones which need more computations. In this thesis,
I supposed that two players of them agree against the
third player by choosing either to cooperate together or
to defect together at each round. According to that assumption,
the game is transformed from the symmetric
three-player model to asymmetric two-player model.Such that, the identities of the players cannot be interchanged
without interchanging the payoff of the strategies.
I determined the payoff matrix corresponding to the
all possible strategies. I noticed that, for some strategies,
it is better to be a player of the first type (independent
player) than being of the second type (allies). In appendix
section, I designed an algorithm and implement it using
the Java programming language to facilitate the calculations.
Key Words:
Iterated games, Prisoner’s dilemma, Payoff matrix, Symmetric
games, Asymmetric games, Tit For Tat strategy,
Evolutionary games.