luweizhang / chess-ai

chess game and AI built with python. (in progress)

Grandmaster Level Chess AI (always in progress)

This is a chess AI and chess game built using python.

The end goal is to have a trained convolutional net be used as an evaluation function to evaluate future states from the tree search. Then, the optimal move will be determined using the minimax algorithm (with alpha beta pruning).

Use python virtual machine to run this:

brew install python@2
virtualenv -p /usr/local/bin/python2 venv 
virtualenv venv
source venv/bin/activate
pip install -r requirements.txt

To play the ai:

python2 play.py

You will be prompted to enter a starting and ending coordinate for your move. I.e a2 -> a4

You can find more details in my blog: luweilikesdata.blogspot.com

Tree generation:

The tree generator component of the chess AI generates the possible future board positions based on the possible moves that the pieces can make. At any given position, there are on average 20 possible moves that can be made in chess.

The above illustrates a chess tree. Possible moves are generated from the current position to create hypothetical future states.

Evaluation function

This is the "intuition" of the chess AI. Given an 8x8 chess position, the evaluation function will determine whether a move is good or not by assigning it a numerical score. A negative score means that black is winning, a positive score means that white is winning.

The evaluation function can take many forms depending on what approach you want to take. You can use hard coded heuristic features. You could take a model based approach (ie. train a neural network), etc etc.

Minimax algorithm:

The minimax algorithm tries to find the optimal move from a tree of moves. Starting from the leaves of the tree which represent future states, you want to propogate back to the present root node by alternating between min() and max() (hence the name, minimax).

When it is your turn, you want to optimize for your outcome, therefore you want to calculate the max() position score.

When it is your opponents turn, he wants to minimize your outcome, therefore it calculates the min() position score.

The guide below explains minimax very well: http://web.cs.ucla.edu/~rosen/161/notes/minimax.html

Alpha-beta pruning:

Alpha-beta pruning is a process that can be used to greatly reduce the space of the tree search. It works by pruning the search tree as you generate it by ruling out paths that neither you nor your opponent will never take.

This is done by keeping track of two values called alpha and beta as you are searching through the tree.
Alpha is the maximum lower bound (the lowest score you are willing to accept)
Beta is the minimum upper bound (the highest score your opponent is willing to accept)

The guide below explains alphabeta pruning very well: http://web.cs.ucla.edu/~rosen/161/notes/alphabeta.html

Appendix

Install Python2 ipython notebook environment

pip install ipython notebook
python2 -m pip install ipykernel #install python2 jupyter kernel
python2 -m ipykernel install --user

chess-ai's Issues

Package Installation Error

ERROR: Could not find a version that satisfies the requirement ipython==7.16.3 (from -r requirements.txt (line 8)) (from versions: 0.10, 0.10.1, 0.10.2, 0.11, 0.12, 0.12.1, 0.13, 0.13.1, 0.13.2, 1.0.0, 1.1.0, 1.2.0, 1.2.1, 2.0.0, 2.1.0, 2.2.0, 2.3.0, 2.3.1, 2.4.0, 2.4.1, 3.0.0, 3.1.0, 3.2.0, 3.2.1, 3.2.2, 3.2.3, 4.0.0b1, 4.0.0, 4.0.1, 4.0.2, 4.0.3, 4.1.0rc1, 4.1.0rc2, 4.1.0, 4.1.1, 4.1.2, 4.2.0, 4.2.1, 5.0.0b1, 5.0.0b2, 5.0.0b3, 5.0.0b4, 5.0.0rc1, 5.0.0, 5.1.0, 5.2.0, 5.2.1, 5.2.2, 5.3.0, 5.4.0, 5.4.1, 5.5.0, 5.6.0, 5.7.0, 5.8.0, 5.9.0, 5.10.0)
ERROR: No matching distribution found for ipython==7.16.3 (from -r requirements.txt (line 8))

Training Data

Hey @luweizhang ,
Great work man !

I have a quick question, Can you please walk me through?

• From where you got the training dataset. (please provide me the dataset)
• Can you please brief how the training is done.