mao-bot

An AI Framework that plays Mao. Lit.

instructions to run

• run "python game.py". make modifications to the exact script your running at the bottom of game.py

to-dos

• Constraints are done.
• Effects are done.
• Need to update RandomAgent and HMM Agent to a) change and b) recognize PoisonDist

Observations -- put things you notice here

• Going second is a big advantage for the learning agents, because they don't have to take a shot in the dark.
• Despite that, LearningAgent still oftentimes outperforms HmmAgent, even with HmmBot's 'second' advantage

architecture

Files & Classes:

• infrastructure.py

  • Card
    • labelled "C", "H", "D" "S" for suit, & 2-14 for values.
  • Deck
    • supports shuffling, resetting, and drawing
  • Rule
    • Simple way to identify rules, without passing an entire Constraint Object

• player.py

  • Player
    • Contains the methods the game queries the player during their turn.
    • read the documentation for details
    • at this point: DO NOT CHANGE

• game.py

  • Game
    • Contains fundamental logic to the game

• agent.py

  • Agent class (inherits from player?)
    • Computation and decision making on top of player class
    • build off of the abstract class you see
    • MOST OF THE WORK NEEDS TO BE DONE HERE!

• Rules

  • IMPLEMENTED RULES:
    • Basic Rules
      • Value Rule: higher or lower values can be played
      • Cards of the same suit can be played, regardless of value
    • Basic Effects
      • Wild Values: a card of this value can always be played. Defaults as "7"
      • Trump Suit: Can Be played at any time (suit)
  • UNIMPLEMENTED RULES:
    • Talking: Choose from a couple of statements every time x card
    • Skip Rules: Certain cards skip the next player.
      • variant 1:
    • Face Cards: cards have effects