View source for Epsilon-equilibrium

{{Infobox equilibrium|
name=Epsilon-equilibrium |
supersetof=[[Nash Equilibrium]]|
usedfor = [[stochastic game]]s
}}

In [[game theory]], an '''epsilon-equilibrium''', or near-Nash equilibrium, is a [[strategy profile]] that approximately
satisfies the condition of [[Nash equilibrium]]. In a Nash equilibrium, no player has an incentive to change his
behavior. In an approximate Nash equilibrium, this requirement is weakened to allow the possibility that a
player may have a small incentive to do something different. This may still be considered an adequate
solution concept, assuming for example [[status quo bias]]. This solution concept may be preferred to Nash
equilibrium due to being easier to compute, or alternatively due to the possibility that in games of more
than 2 players, the probabilities involved in an exact Nash equilibrium need not be [[rational numbers]].<ref>{{cite journal
| author = V. Bubelis
| year = 1979
| title = On equilibria in finite games
| journal = International Journal of Game Theory
| volume = 8
| issue = 2
| pages = 65–79
| doi=10.1007/bf01768703
| s2cid = 122843303
}}</ref>

== Definition ==

There is more than one alternative definition.

=== The standard definition ===

Given a game and a real non-negative parameter <math>\varepsilon</math>, a [[strategy profile]] is said to be an
<math>\varepsilon</math>-equilibrium if it is not possible for any player to gain more than <math>\varepsilon</math> in [[payoff dominant equilibrium|expected payoff]] by unilaterally deviating from his [[Strategy (game theory)|strategy]].<ref name=AGT>{{Cite Algorithmic Game Theory 2007}}</ref>{{rp|45}} 
Every [[Nash Equilibrium]] is equivalent to an <math>\varepsilon</math>-equilibrium where <math>\varepsilon = 0</math>.

Formally, let <math>G = (N, A=A_1 \times \dotsb \times A_N, u\colon A \to R^N)</math>
be an <math>N</math>-player game with action sets <math>A_i</math> for each player <math>i</math> and utility function <math>u</math>.
Let <math>u_i (s)</math> denote the payoff to player <math>i</math> when [[Strategy (game theory)|strategy profile]] <math>s</math> is played.
Let <math>\Delta_i</math> be the space of probability distributions over <math>A_i</math>.
A vector of strategies <math>\sigma \in \Delta = \Delta_1 \times \dotsb \times \Delta_N</math> is an <math>\varepsilon</math>-Nash Equilibrium for <math>G</math> if 
:<math>u_i(\sigma)\geq u_i(\sigma_i^',\sigma_{-i})-\varepsilon</math> for all <math>\sigma_i^' \in \Delta_i, i \in N.</math>

=== Well-supported approximate equilibrium ===

The following definition<ref>{{cite conference
| doi = 10.1145/1132516.1132526
| author = P.W. Goldberg and [[Christos Papadimitriou|C.H. Papadimitriou]]
| year = 2006
| title = Reducibility Among Equilibrium Problems
| book-title = 38th Symposium on Theory of Computing
| pages = 61–70
}}</ref>
imposes the stronger requirement that a player may only assign positive probability to a pure strategy <math>a</math> if the payoff of <math>a</math> has expected payoff at most <math>\varepsilon</math> less than the best response payoff.
Let <math>x_s</math> be the probability that strategy profile <math>s</math> is played. For player <math>p</math> let <math>S_{-p}</math> be strategy profiles of players other than <math>p</math>; for <math>s\in S_{-p}</math> and a pure strategy <math>j</math> of <math>p</math> let <math>js</math> be the strategy profile where <math>p</math> plays <math>j</math> and other players play <math>s</math>.
Let <math>u_p(s)</math> be the payoff to <math>p</math> when strategy profile <math>s</math> is used.
The requirement can be expressed by the formula
:<math>\sum_{s\in S_{-p}}u_p(js)x_s > \varepsilon+\sum_{s\in S_{-p}}u_p(j's)x_s \Longrightarrow x^p_{j'} = 0.</math>

== Results ==

The existence of a [[polynomial-time approximation scheme]] (PTAS) for &epsilon;-Nash equilibria is
equivalent to the question of whether there exists one for &epsilon;-well-supported
approximate Nash equilibria,<ref>{{cite journal
| doi = 10.1137/070699652
| author = C. Daskalakis, P.W. Goldberg and [[Christos Papadimitriou|C.H. Papadimitriou]]
| year = 2009
| title = The Complexity of Computing a Nash Equilibrium
| journal = [[SICOMP|SIAM Journal on Computing]]
| volume = 39
| issue = 3
| pages = 195–259
| citeseerx = 10.1.1.68.6111
}}</ref> but the existence of a PTAS remains an open problem.
For constant values of &epsilon;, polynomial-time algorithms for approximate equilibria
are known for lower values of &epsilon; than are known for well-supported
approximate equilibria.
For games with payoffs in the range [0,1] and &epsilon;=0.3393, &epsilon;-Nash equilibria can
be computed in polynomial time.<ref>{{cite journal
| author = H. Tsaknakis and Paul G. Spirakis
| year = 2008
| title = An optimisation approach for approximate Nash equilibria
| journal = Internet Mathematics
| volume = 5
| issue = 4
| pages = 365–382
| doi=10.1080/15427951.2008.10129172
| doi-access = free
}}</ref>
For games with payoffs in the range [0,1] and &epsilon;=2/3, &epsilon;-well-supported equilibria can
be computed in polynomial time.<ref>{{cite journal
| doi = 10.1007/s00453-008-9227-6
| author = Spyros C. Kontogiannis and Paul G. Spirakis
| year = 2010
| title = Well Supported Approximate Equilibria in Bimatrix Games
| journal = [[Algorithmica]]
| volume = 57
| issue = 4
| pages = 653–667
| s2cid = 15968419
}}</ref>

== Example ==

The notion of ε-equilibria is important in the theory of 
[[stochastic game]]s of potentially infinite duration. There are 
simple examples of stochastic games with no [[Nash equilibrium]] 
but with an ε-equilibrium for any ε strictly bigger than 0.

Perhaps the simplest such example is the following variant of [[Matching Pennies]], suggested by Everett. Player 1 hides a penny and
Player 2 must guess if it is heads up or tails up. If Player 2 guesses correctly, he
wins the penny from Player 1 and the game ends. If Player 2 incorrectly guesses that the penny 
is heads up,
the game ends with payoff zero to both players. If he incorrectly guesses that it is tails up, the game '''repeats'''. If the play continues forever, the payoff to both players is zero.

Given a parameter ''ε'' > 0, any [[strategy profile]] where Player 2 guesses heads up with
probability ε and tails up with probability 1&nbsp;&minus;&nbsp;''ε'' (at every stage of the game, and independently
from previous stages) is an ''ε''-equilibrium for the game. The expected payoff of Player 2 in
such a strategy profile is at least 1&nbsp;&minus;&nbsp;''ε''. However, it is easy to see that there is no
strategy for Player 2 that can guarantee an expected payoff of exactly 1. Therefore, the game
has no [[Nash equilibrium]].

Another simple example is the finitely [[Repeated game#Repeated prisoner.27s dilemma|repeated prisoner's dilemma]] for T periods, where the payoff is averaged over the T periods. The only [[Nash equilibrium]] of this game is to choose Defect in each period. Now consider the two strategies [[tit-for-tat]] and [[grim trigger]]. Although neither [[tit-for-tat]] nor [[grim trigger]] are Nash equilibria for the game, both of them are <math>\epsilon</math>-equilibria for some positive <math>\epsilon</math>. The acceptable values of <math>\epsilon</math> depend on the payoffs of the constituent game and on the number T of periods.

In economics, the concept of a [[pure strategy]] '''epsilon-equilibrium''' is used when the mixed-strategy approach is seen as unrealistic. In a pure-strategy epsilon-equilibrium, each player chooses a pure-strategy that is within epsilon of its best pure-strategy. For example, in the [[Bertrand–Edgeworth model]], where no pure-strategy equilibrium exists, a pure-strategy epsilon equilibrium may exist.

== References ==
;Inline citations
{{Reflist}}

;Sources
*[[Huw Dixon|H Dixon]] [https://ideas.repec.org/a/bla/restud/v54y1987i1p47-62.html Approximate Bertrand Equilibrium in a Replicated Industry], Review of Economic Studies, 54 (1987), pages 47–62.
*H. Everett. "Recursive Games". In H.W. Kuhn and A.W. Tucker, editors.  ''Contributions to the theory of games'', vol. III, volume 39 of ''Annals of Mathematical Studies''. Princeton University Press, 1957.
* {{Citation | last2=Shoham | first2=Yoav | last1=Leyton-Brown | first1=Kevin | authorlink1 = Kevin Leyton-Brown | title=Essentials of Game Theory: A Concise, Multidisciplinary Introduction  | publisher=Morgan & Claypool Publishers | isbn=978-1-59829-593-1 | url=http://www.gtessentials.org | year=2008 | location=San Rafael, CA}}. An 88-page mathematical introduction; see Section 3.7. [http://www.morganclaypool.com/doi/abs/10.2200/S00108ED1V01Y200802AIM003 Free online] {{Webarchive|url=https://web.archive.org/web/20000815223335/http://www.economics.harvard.edu/~aroth/alroth.html |date=2000-08-15 }} at many universities.
*[[Roy Radner|R. Radner]]. ''Collusive behavior in non-cooperative epsilon equilibria of oligopolies with long but finite lives'', Journal of Economic Theory, '''22''', 121–157, 1980.
* {{Citation | last1=Shoham | first1=Yoav | last2=Leyton-Brown | first2=Kevin | title=Multiagent Systems: Algorithmic, Game-Theoretic, and Logical Foundations | publisher=[[Cambridge University Press]] | isbn=978-0-521-89943-7 | url=http://www.masfoundations.org | year=2009 | location=New York}}. A comprehensive reference from a computational perspective; see Section 3.4.7. [http://www.masfoundations.org/download.html Downloadable free online]. 
*S.H. Tijs. ''Nash equilibria for noncooperative ''n''-person games in normal form'', SIAM Review, '''23''', 225–237, 1981.

{{Game theory}}

{{DEFAULTSORT:Epsilon-Equilibrium}}
[[Category:Game theory equilibrium concepts]]