BlackJack card game in SWI-Prolog

Introduction

Implementation of a BlackJack card game in prolog and r..eal library. The later is an interface to R statistical package within SWI code.

To run the code one needs to have R, R-dev, SWI-Prolog and r..eal installed.

The pldoc generated documentation of the code is available here.

Pre-configuration

There are number of parameters available, to customize the program. They are listed at the top of each package file.

1. Plays - Number of games to be played. (blackJack.pl)

2. Decks - Number of decks used in a single game. (deck.pl)

3. Players - Number of AI players. AI1 plays deterministic table, AI2 plays deck probabilities, AI3 plays probabilistic choice, AI4 plays hold on all. If not otherwise defined, all other players are using deterministic table. (blackJack.pl)

4. PlayerMode - The game-play can be either defined to be: experimental - no user interaction required, this mode is mainly used to test AI algorithms; or interactive - user can play along AI algorithms. (blackJack.pl)

5. ShuffleMode - The simulation performs riffle shuffle(see the report); it can be either deterministic - there is only one (fair)coin toss to decide whether left or right pile starts the shuffle; or random - coin is tossed for each card(from left pile) to decide whether it goes on the top or on the bottom. (deck.pl)

6. InitShuffles - Number of deck shuffles before the game starts. (blackJack.pl)

7. Shuffles - Number of deck shuffles between games. (blackJack.pl)

8. Dealer - Dealer strategy to be used: H17 and S17, both described in details in the report. (dealerStrategies.pl)

Executing the program(i.e. playing BlackJack)

To run the script:

 cd ~
 git clone [email protected]:So-Cool/BlackJack.git
 cd BlackJack
 swipl

and then within SWI-Prolog:

[blackJack].
blackJack.

After the game is finished there will be a score graph and score table(in the .csv format) produced within the BlackJack directory. The rows in the file are scores from particular round and the columns are in order: dealer, AI1-deterministic table, AI2-deck probabilities, AI3-mixture of previous two, AI4-always stand and finally, if enabled, user's score. The same order applies to vertical bars visible in the plot.

Remark on point counting

The scores are win/lose based; the player score is only compared against the dealer's points:

• if the player has more points than the dealer and has not exceeded 21 only he gets a point;
• if the player has less points than the dealer and has not exceeded 21 only the dealer gets a point;
• if there is a draw between the player and the dealer and both have less than 21 points both get a point;
• if there is a draw between the player and the dealer and both have more than 21 points no one get a point;
• if the player exceeded 21 points and the dealer has not only the dealer gets a point;
• if the dealer exceeded 21 points and the player has not only the player gets a point.

---

Warning! it may become addictive Warning!

---

For more information please refer to the report available within this repository.

---

ToDo

• Possibility to split on pairs.
• Strategy idea: create a sampling distribution over all possible cards that can be drawn and sample a card ,based on that card predict according to table whether it is worth taking it or not.
• Preserve the deck after each game and shuffle it instead of discarding (adds more variability):

%% define number of shuffles after each game
shuffles(2).