Minimax

A game of perfect knowledge

Perfect knowledge means all information is available to all the players. For example in chess, both players can see all pieces and where they are. In a game like poker, there is hidden information. You can't see what's in the deck and you can't see what is in other players hands. Minimax is nice because it has perfect knowledge without being too computationally expensive, so it works well as a demonstration.

How to play

Game play is very simple in Minimax. To start you need to draw yourself a perfect binary tree. Then assign each leaf in the tree a random number (or "random" if you wann rig the game). Starting from the top of the tree, 2 players now take turns navigating down the tree. When you reach the bottom of the tree, the value at that leaf is the result of the game. The goals for each player are polar opposite. One player wishes to maximize the final value while the other wishes to minimize it. I'm not sure there is a winner or a loser in this game, but if you wanna develop your own method for determining a winner and and a loser then be my guest.

How the program works

The program is just a command line version of minimax. Once you run the program you'll be prompted to enter a command. The command list is as follows:

example

Walks you through an example game

help

Brings up a help menu with a list of commands

begin

Begins the game

set {key} = {value}

Example: set depth = 6
Sets some paramaters for the game. Currently the available paramaters are:

• singleplayer - A boolean value which which sets whether you play against the computer or another person
• depth - An integer value which sets the depth of the tree
• min - One of either "player_one" OR "player_two". Sets that player's goal to minimize the value
• max - One of either "player_one" OR "player_two". Sets that player's goal to maximize the value

quit

Exits the program

Playing the game

Once you enter the begin command, a tree will be generated and the game will begin. Players take turns entering either left or right and when the end is reached the value will be announced. You can see where the you currently are in the tree by looking for the caret ^ underneath the branch point. You can also look for the right angle bracket > next to the line numbers on the left to see which row you are currently in.