What is the difference between correlated equilibrium and mixed equilibrium?
Here's what I understand :
Unlike a pure Nash equilibrium, a mixed equilibrium corresponds to when each player has a probability distribution he follows instead of pure actions (single action with probability one).
Also, what I understand by correlated equilibrium is that it corresponds to when someone tells each player to follow some particular probability distribution, and each player plays according to that.
What I don't understand is what is the difference between these two? Can't one be modeled as the other?
$\endgroup$1 Answer
$\begingroup$In a mixed (strategy) Nash equilibrium, the players' actions (strategies) are independent random variables. In other words, if you know that player 1 (randomly) chose $x$, that doesn't give you any additional information about what player 2 might do.
In correlated equilibria the actions (strategies) need not be independent. Wikipedia gives an example in the setting of the game of chicken, showing that the expected value of a correlated equilibrium can be greater than that of the mixed strategy equilibrium.
$\endgroup$ 2
