Mixed Equilibria


In this chapter, we expand the range of possible actions that a player can take. Rather than committing to a single action (in a pure strategy), players can now use mixed strategies where they assign a probability to each action. Essentially, they commit to a probabilistic distribution of actions.

We can formalize this concept of a mixed strategy with some notation:

  • Let $\alpha_i$ (an alpha, not an a) represent a mixed strategy profile.
  • Let $\alpha_i(a_i)$ represent player $i$’s mixed strategy, a function that assigns a probability to each pure strategy $a_i.$
    • For convenience like before, $\alpha_{-i}$ denotes ever other player’s mixed strategy.
    • For convenience, we can express $\alpha_i$ as an ordered list of probabilities that correspond to actions.
  • Let $U_i(\alpha)$ (with a capital U) represent the expected payoff, or Bernoulli payoff function for player $i$ of the mixed strategy profile $\alpha.$ Symbolically, it is defined as $$U_i(\alpha) = \sum_{a_i \in A_i} \alpha_i (a_i) E(a_i, a_{-i}),$$ where $E(a_i, \alpha_{-i})$ represents the expected payoff for player $i$ if they play pure strategy $a_i$ and everyone else plays the mixed profile $\alpha_{-i}.$
These variable names are somewhat standard and will be followed throughout the website.
In the coming chapters, we explore some scenarios where mixed equilibria arise to motivate this unintuitive concept.
Next Chapter: An Impossible Choice