A re-implementation of 3D Tetris, along with a genetic programming algorithm that steadily improves its performance of playing the game.
Includes visualizations through a chart and replays of "fit" individuals in each generation.
Node is required to run the simulation. Python 2 is required to run the genetic algorithm. Python 3 is required to run the web server
-
Python 2.7 or higher (for genetic algorithm), and also python-pip
-
Python 3.5 or higher (for web server), and also python3-pip
-
Node 6.x or higher (for simulation)
-
Clone the repository and change directory to it.
-
Set up Python virtual environment. Also make sure that your working directory has no spaces in it.
python3 -m pip install virtualenv flask
python3 -m virtualenv env
python -m pip install virtualenv enum zerorpc
python -m virtualenv env2- Set up NPM dependencies.
cd simulation
npm installTo go into a virtual environment: (Linux/Mac) source env/bin/activate (Windows) env\Scripts\activate
To run, you need two terminals.
-
cd simulation && node index.js -
(Make sure this one is in the virtual environment
env2)python genetic/genetic_algorithm.py
Genetic algorithm will run until 1000 generations by default. This can be configured in genetic_algorithm.py.
- (Make sure this one is in the virtual environment
env)python3 genetic/webserver.py
Server will be visible, by default, on port 80 on localhost. You can compare it to our web server for reference.
The web server hosts data files which consist of a csv file (generations vs score) and a json file (replay data) which configure the ai endpoint. The / endpoint lets you play the game on a browser with WebGL and ES6 support.
Pull requests will always be welcome!
[ ] Use a transpiler for older browser support [ ] Improve genetic algorithm [ ] Alternate views for the 3D Tetris game [ ] Improve documentation [ ] Add testing [ ] And more...