2013年10月9日 星期三

顫抖手精煉均衡 (trembling hand perfect equilibrium)


Definitioin

“顫抖手精煉均衡”概念是澤爾騰提出的對納什均衡的一個改進。顫抖手精煉均衡的基本思想是:在任何一個博弈中,每個局中人都有一定的犯錯誤的可能性(類似一個人用手抓東西時,手一顫抖,他就抓不住他想抓的東西)。一個策略對是一個顫抖手精煉均衡時,它必須具有如下性質:各局中人i要採用的策略,不僅在其他局中人不犯錯誤時是最優的,而且在其他局中人偶爾犯錯誤(概率很小,但大於0)時還是最優的。可以看出,顫抖手精煉均衡是一種較穩定的均衡。

all text is from MBALib:
http://wiki.mbalib.com/zh-tw/%E9%A2%A4%E6%8A%96%E6%89%8B%E7%B2%BE%E7%82%BC%E5%9D%87%E8%A1%A1




Example

The game represented in the following normal form matrix has two pure strategy Nash equilibria, namely <Up, Left> and <Down, Right>. However, only <U,L> is trembling-hand perfect.
LeftRight
Up1, 12, 0
Down0, 22, 2
Trembling hand perfect equilibrium
Assume player 1 is playing a mixed strategy (1-\epsilon, \epsilon), for  0<\epsilon <1. Player 2's expected payoff from playing L is:
1(1-\epsilon) + 2\epsilon = 1+\epsilon.\
Player 2's expected payoff from playing the strategy R is:
0(1-\epsilon) + 2\epsilon = 2\epsilon.\
For small values of ε, player 2 maximizes his expected payoff by placing a minimal weight on R and maximal weight on L. By symmetry, player 1 should place a minimal weight on D if player 2 is playing the mixed strategy (1-\epsilon, \epsilon). Hence <U,L> is trembling-hand perfect.
However, similar analysis fails for the strategy profile <D,R>.
Assume player 2 is playing a mixed strategy (\epsilon, 1-\epsilon). Player 1's expected payoff from playing U is:
1\epsilon + 2(1-\epsilon) = 2-\epsilon.\
Player 1's expected payoff from playing D is:
0(\epsilon) + 2(1-\epsilon) = 2-2\epsilon.\
For all positive values of ε, player 1 maximizes his expected payoff by placing a minimal weight on D and maximal weight on U. Hence <D, R> is not trembling-hand perfect because player 2 (and, by symmetry, player 1) maximizes his expected payoff by deviating most often to L if there is a small chance of error in the behavior of player 1.

all text is from Wikipedia
http://en.wikipedia.org/wiki/Trembling_hand_perfect_equilibrium