Graphical game theory: Difference between revisions

Revision as of 06:01, 28 January 2010

Revision as of 06:12, 28 January 2010

Line 1:

In [[game theory]], The common ways to describe a game are the [[normal-form game|normal-form]] and the [[extensive-form game|extensive form]]. The graphical form is an alternate way to describe a game in a compact representation using the interaction among participants.

#REDIRECT [[Video game]]

Consider a game with <math>n</math> players with <math>m</math> strategies each. We will represent the players as nodes in a graph in which each player has a utility function that depends only on him and his neighbors. As the utility function depends on fewer other players, the graphical representation would be smaller.

==Formal definition==

A graphical game is represented by a graph <math>G</math>, in which Each player is represented by a node, and there is an edge between two nodes <math>i,j</math> iff their utility functions are depended on the strategy which the other player will choose . Each node <math>i</math> in <math>G</math> has a function <math>u_{i}:\{1\ldots m\}^{d_{i}+1}\rightarrow\mathbb{R}</math>, where <math>d_i</math> is the degree of vertex <math>i</math>. <math>u_{i}</math> specifies the utility of player <math>i</math> as a function of his strategy as well as those of his neighbors.

==The size of the game's representation==

For a general <math>n</math> players game, in which each player has <math>m</math> possible strategies, the size of a normal form representation would be <math>O(m^{n})</math>. The size of the graphical representation for this game is <math>O(m^{d})</math> where <math>d</math> is the maximal node degree in the graph. If <math>d\ll n</math>, then the graphical game representation is much smaller.

==An example==

In case where each player's utility function depends only on one other player, the maximal degree of the graph is 1, and the game can be described as <math>n</math> functions (tables) of size <math>m^{2}</math>. So, the total size of the input will be <math>nm^{2}</math>.

==Nash Equilibrium==

Finding Nash equilibrium in a game takes exponential time in the size of the representation. If the graphical representation of the game is a tree, we can find the equilibrium in polynomial time. In the general case, where the maximal degree of a node is 3 or more, the problem is [[NP-complete]].

[[he:משחק בצורה גרפית]]