Fullway handout
-
Finitely repeated games
We ‘repeat’ a normal form game \(T\) times. Strategies need to be defined to take in to account the entire history of the game.
What did we do?
- We played an iterated prisoner’s dilemma tournament.
What do I need to be able to do?
- Define a repeated game.
- State and prove the theorem of a sequence of stage Nash profiles.
- Draw possibly utility outcomes.
- Obtain subgame perfect Nash equilibria that are not a sequence of state Nash equilibria.
-
Infinitely repeated games
We now ‘repeat’ a normal form game an infinite amount of times. To deal with this we use discounting.
What did we do?
- We played an infinite iterated prisoner’s dilemma tournament by introducing a probability of the game carrying on.
What do I need to be able to do?
- Calculate expected utilities with discounting.
- Prove particular instances where a given strategy is better than another.
- Prove the Folk theorem.
-
Population Games and Evolutionary stable strategies
We move on to something different and consider populations and how a particular strategy evolves in a population.
What did we do?
- We played a brief game with cards in class.
What do I need to be able to do?
- Define a population game.
- State and prove the theorem for necessity of stability.
- Define a post entry population \(\chi_{\epsilon}\) and an ESS.
- Obtain ESS in a game against the field.
-
Nash equilibrium and Evolutionary stable strategies
We consider a particular type of game against the field: member of the population are randomly matched and play a normal form game.
What did we do?
- Nothing really.
What do I need to be able to do?
- Define a pairwise contest game.
- Find ESS in a pairwise contest game from first principles.
- State prove and use the theorem relating an ESS to a Nash equilibrium of a normal form game.
-
Random events and incomplete information
Here we look at extensive form games with random events.
What did we do?
- 3 volunteers played a version of matching pennies with some randomness involved (someone won a melon)
What do I need to be able to do?
- Obtain the normal form representation of a game with incomplete information.
- Describe basic utility theory.
- Solve the principal agent game.
-
Stochastic games
Games where strategies not only define the utilities but the next game.
What did we do?
- We played a modified version of the Prisoner’s dilemma in the form of a stochastic game (this was a 4 team round robin tournament)
What do I need to be able to do?
- Define a stochastic/Markov game, a Markov strategy.
- State the requirement for Nash equilibrium.
- Obtain Nash equilibrium for stochastic games.
-
Matching games
We look at the stable marriage problem and the Gale Shapley algorithm.
What did we do?
- We matched 3 volunteers to 3 of my toys (Donatello, Zoe and my Tech Deck).
What do I need to be able to do?
- Define a matching game.
- State and apply the Gale-Shapley algorithm.
- State and prove the theorem guaranteeing a unique matching as output of the Gale Shapley algorithm.
- State and prove the theorem of reviewer sub optimality.
-
Cooperative games
Sharing resources when considering contribution of coalitions.
What did we do?
- We played basketball (kind of).
What do I need to be able to do?
- Define a characteristic value game.
- Define monotone and superadditive games.
- Define Efficiency, Null player, Symmetry and Additivity (see exercises - you could be asked to prove that the Shapley value has these properties).
- Define and be able to calculate the Shapley value.
-
Routing games
Games that can be used to model behaviour in system affected by congestion.
What did we do?
- Nothing fun at all…
What do I need to be able to do?
- Define a routing game, set of paths and feasible paths.
- Define and obtain (from first principles) Optimal flows.
- Define and obtain (from first principles) Nash flows.
- Define the potential function.
- Define marginal costs.
- State and use the theorem connecting the Nash flow to the optimal flow.
- State and use the theorem connecting optimal flows to Nash flows.