NAME : Sumedh Vilas Datar
LANUGUAGE : Python
As part of the assignment I have implemented minimax, alpha beta pruning and depth limited minimax for prediciting the next best move in connect4 game while playing against a human or a bot.
I have implemented the game in two modes which are interactive mode and one-move mode.
The game uses two python files called maxconnect4.py and MaxConnect4Game.py.
maxconnect4.py has all the functions related to setting up of the game like taking inputs from the user and setting up the board and calling the ai algorithms.
MaxConnect4Game.py has all the functions which involves implementing the above ai game algorithms and implementation of the evaluation function.
All algorithm implementation is done in a method called aiPlay() between lines 270 and 272. Default it is using minimax with alphabeta pruning and other methods are commented. Other methods can be used by uncommenting.
Sr No | Method | Operation |
- | self.minimax() | Plain Minimax algorithm implemented |
- | self.alpha_beta_decision | Minimax with alpha beta pruning |
- | self.depth_limited_alpha_beta_pruning | Depth limited minimax with alpha beta pruning and specified depth |
Interactive Mode Syntax :
$python maxconnect4.py interactive input.txt computer-next/human-next depth The above code will allow user to give his input and simulate the connect 4 game till all 42 moves are made and it will tell the final result as to who won and who lost.
One-move Mode Syntax :
$python maxconnect4.py one-move input_file output_file depth The above code will predict the best possible next move and make the move given an input state.
Partly implemented the evaluation function from what professor explained in class and partly I took the idea from the following link below to implement the evaluation function
https://github.com/erikackermann/Connect-Four
Evaluation function for depth limited minimax is implemented to calculate the utility value.
Calculation of Utility Value from the evaluation function :
The higher the utility value then better is the decision of the computer in selecting the respective column.
First it checks for whether it can make a four, then it checks subsequently for threes and then twos. Similarly it checks for the highest possible four of the opponent and the formuala is given below :-
utility_value = (my_fours * 10 + my_threes * 5 + my_twos * 2)- (opp_fours *10 + opp_threes * 5 + opp_twos * 2). After calculating all utility values, the highest value and the corresponding column mapped to it is selected and then the move is predicted.
An excel sheet named DepthVsTime has a table containing the time taken and the corresponding depth level. Time is calculated for the corresponding depth till time reaches 1 minute and then it is stopped.
After conducting the above experiment I could observe that time will cross 1 minute at depth 8 and beyond.