Skip to content

Latest commit

 

History

History
41 lines (33 loc) · 2.96 KB

README.md

File metadata and controls

41 lines (33 loc) · 2.96 KB

F The Dealer Simulation

The Game

F the Dealer is a drinking game with poker playing cards. One player starts as the dealer and holds the deck of cards. The player next to him now has to guess, what number lies on top of the stack. If he's correct, good, if not, he gets a second chance: The dealer now tells the current guesser if the number of the correct card is above or below his first guess. He gets to guess again. Now depending on his guesses they have to drink:

  • The dealer 5 times if the guesser was correct on first try.
  • The dealer 2 times if the guesser was correct on second try.
  • The guesser 2 times if he was wrong on both tries.

After that each card is put on the table for everyone to see and current guesser changes to the next player. The dealer doesn't change until two guessers are wrong in a row.

Motivation to simulate

Of course it would be best to calculate probabilities for each card at each turn, to get the best result. But because that's not a doable strategy in real life, there is the question whether one can find an easy rule of thumb to help choosing a card. The simulation focuses on comparing different strategies by counting how often one guesses right/wrong or has to drink.

Insight in the simulation

The main class is Game, it contains the game's logic and keeps track of the scores with a Score object. The Player class is an abstract super class and defines which actions a player must have implemented, see Player type tree below. The Deck class holds a representation (optimized) of a deck of cards. A player can post requests to it (e.g. how many cards there are etc.).

Running

For a single run one can call run() in the Game class. To do multiple runs and comparing how good the current chosen Players are, one can use runAndResetNTimes(). Also good for debugging to see if a player is about as good as expected.

Output and generating game data

The current purpose of the main method in the Game class is a call to generateDataset() which runs lots of rounds and saves them in a .csv file for each player.

Player type tree

There are several playertypes (strategies) that can be chosen for the simulation. (Feel free to add other types!) Italic means abstract class.

  • Player: General Player class. Defines header of firstGuess() and secondGuess().
    • ProbabilisticPlayer: Can calculate various probabilities. Mean't to be expanded. Returns at first guess the card with the highest value given by value() function.
      • HighestProbabilityPlayer: Values the probability of success for each card, considering the first and second guess.
      • HighestProbabilityOnFirstTryPlayer: Values the probability of success for each card, considering only the first guess.
      • HighestExpectationValuePlayer: Values the expectation value for each card, considering the first and second guess.
    • RandomPlayer: Always guesses a random (but still available) number. Good for comparing to a new class for testing.