Hearts

How To Run

Using Command Line:

1. Compile all of the source files by running javac *.java or ant
2. Run the Hearts program by running java Hearts

Using an IDE

1. Compile all of the source files using your IDE
2. Run the compiled Hearts program using your IDE

Implementing and Adding Your Own Agents

To implement your own AI agent, all you need to do is write a new class that extends Player. As an example, you can take a look at the existing classes such as LookAheadPlayer or RandomPlayAI.

Once they are implemented, you can add them to the game by modifying p1, p2, p3, or p4 in Hearts.java to instantiate a new agent of your class.

Implemented Players

• HumanPlayer: Allows humans to play Hearts
• LowPlayAI: Picks the lowest card in a valid suit
• RandomPlayAI: Plays a random card in a valid suit
• LookAheadPlayer: Looks ahead a minimal amount of steps to make a decision
• MCTSPlayer: Uses MCTS to decide on the optimal next move

Class Structure

• Value enum
• Suit enum
• Card class (for each individual card)
• Deck class (for the "deck" to deal hands)
• Hand class (for each of the hands)
• Player abstract class (defines base functionality for any player)
• HumanPlayer class extends Player (performAction() allows human input)
• State class (track state of the game and allow for random playouts)
• Game class (plays one game)
• Hearts class (this is the main file that brings it all together)

Version History

0.0.1 — Setting all of the infrastructure up and testing some java things

0.0.2 — Implemented abstract player as well as basic human player skeleton

0.0.3 — Begun implementation for game playouts

0.0.4 — Able to do multiple playouts of the game

0.0.5 — Able to keep track of cards played for each round

0.0.6 — Game keeps track of the scores of each player now!

0.0.7 — Game keeps track of scores between games; games do initialization properly now

0.0.8 — HumanPlayer can now select which card to play, with Exception Handling!

0.0.9 — Added card "shorthand" display using unicode characters (and reordered suits)

0.0.10 — Game now checks if the played card by the Player is valid! (for ALL cases)

0.0.11 — Game now checks who the "winner" of each round is!

0.1.0 — First full game implemented! (With no state tracking functionality for AI)

0.2.0 — Game now passes in currentRound to performAction()

0.3.0 — Working Low-Play AI has been completed

0.4.0 — Added score-tracking statistics

0.5.0 — Shooting the moon is implemented — minor bug fixes as well

0.6.0 — Added a copy constructor for the Deck and began working on the State class

0.6.1 — All Players now take parameters of the State class, State constructor now finished

0.6.2 — Added functionality to return the range of cards in any given suit

0.6.3 — RandomPlayAI has been implemented completely and is working

0.6.4 — MCTSPlayer implementation has begun

0.6.5 — LookAheadPlayer implementation has begun

0.7.0 — Implementation for advance() within the State class has begun

0.7.1 — advance() now checks for validity of the selected move

0.7.2 — Design paradigm for multiple game playouts has been established

0.8.0 — Finished preliminary version of advance() in the State class

0.8.1 — Added functionality for displaying debug messages only for Human players

0.8.2 — Fixed printout issue with shooting the moon — random AI able to shoot the moon!

0.9.0 — LookAheadPlayer has performAction() implemented (but not playoutGame())

0.9.1 — Fixed shallow copy issues with Decks and hands

0.9.2 — Playthroughs with state and lookahead player both work! Fixed minor issues

1.0.0 — Full working game, with working State class, game playthroughs, and LookAhead AI

1.0.1 — Finished skeleton of MCTS Player

1.0.2 — Fixed bug with advance() method in the State class with returning accrued points

1.0.3 — Most MCTS functions have been written; just need to deal with random playouts

1.0.4 — Debugging process for MCTS — fixed an issue with treePolicy()

1.0.5 — All MCTS functions except assignReward() now are fully functional

1.0.6 — Preliminary completion of assignReward() function

1.0.7 — MCTS completed and running — checking for validity of results now

1.1.0 — MCTS player completed (but not optimized) for now

Future Work

1. Reorganize file structure
2. Create debug flags for running games without any output (except final win statistics)
3. Decouple Hand and Player operations by creating a new Hand class
4. Fix how cards are removed from the hand/selected by Players