Dynamic Connect-4

This is a Python implementation of an agent that can play the Dynamic Connect-4 game. This is assignment 1 of the ECSE-526 class, as described here.

The agent uses a flavour of alpha-beta pruning with transposition table and heuristic function to search the state space and pick the best move.

Play

To play the game, use the command-line interface in main.py. To see the list of available commands, run the following:

python main.py --help

This will print the following, summarizing the ways of playing or watching the game:

usage: main.py [-h]
               {human_vs_human,human_vs_ai,ai_vs_ai,ai_vs_server,human_vs_server}
               ...

Dynamic Connect-4. To play or watch a game, use one of the positional
arguments.

positional arguments:
  {human_vs_human,human_vs_ai,ai_vs_ai,ai_vs_server,human_vs_server}
    human_vs_human      Play as a human versus another human.
    human_vs_ai         Play as a human versus an AI.
    ai_vs_ai            Spectate an AI versus AI game.
    ai_vs_server        Spectate an AI versus a player on a server.
    human_vs_server     Play as a human versus a player on a server.

optional arguments:
  -h, --help            show this help message and exit

Optional Arguments

State

To manually specify a state, use the following argument:

  -s STATE, --state STATE
                        The name of the file containing the initial state of
                        the game.

Note that the command expects a file with state in the following format:

 , , , , , ,X
X, , , , , ,O
O, , , , , ,X
X, , , , , ,O
O, , , , , ,X
X, , , , , ,O
O, , , , , ,

Note that if the initial state is not specified, the program assumes that there is a states directory (at the same level as main.py) containing an initial_state.txt file.

Colour

When playing any game mode besides human_vs_human or ai_vs_ai, your colour must be specified.

  -c COLOUR, --colour COLOUR
                        Your colour.

Time Limit

When playing with a local AI agent (with human_vs_ai, ai_vs_ai or ai_vs_server), the time limit for a move can be set:

  -t TIME_LIMIT, --time_limit TIME_LIMIT
                        The time limit for a move, in seconds.

Remote Play Arguments

When playing remotely (either with ai_vs_server or human_vs_server), there is the following set of optional arguments:

  -H HOST, --host HOST  Server host address.
  -p PORT, --port PORT  Port number.
  -g GAME_ID, --game_id GAME_ID
                        Game ID.

Default Values

All of the optional arguments above have default values, if none are provided by the user. Here are these default values:

Argument	Default Value
--colour	white
--state	states/initial_state.txt
--time_limit	19
--host	localhost
--port	12345
--game_id	game_id

A sample log of the output of the program (when using the ai_vs_ai mode) can be seen in logs/sample_log.txt.

Example Commands

Here are some example commands:

python main.py ai_vs_server -H localhost -p 12345 -g game_id_123 -t 19 -c white

python main.py ai_vs_ai -t 10

python main.py human_vs_human -c white

Code Organization

There are four main Python files that contain the bulk of the program code, outlined in the following table:

File	Contents
connect_four.py	State, board and action information.
heuristics.py	Heuristics tested or used by the program.
search.py	Search methods, including minimax, negamax and iterative deepening search.
main.py	Main method to parse command-line arguments and execute the game.

The files with prefix test were used to test various aspects of the game, and the graph_creator.py script was used to create the graphs for the assignment report, alongside the MATLAB scripts in the matlab directory.