Conway's Game of Life in Tensorflow

Conway's Game of Life in Tensorflow

View on Github ~~>
cellular automata
live demo

This is a reimplementation of Conway’s Game of Life using TensorFlowJS. The game is a zero-player game, meaning that its evolution is determined by its initial state, requiring no further input. One interacts with the Game of Life by creating an initial configuration and observing how it evolves.

The rules are simple:

  • Any live cell with fewer than two live neighbours dies, as if by underpopulation.
  • Any live cell with two or three live neighbours lives on to the next generation.
  • Any live cell with more than three live neighbours dies, as if by overpopulation.
  • Any dead cell with exactly three live neighbours becomes a live cell, as if by reproduction.


The game is implemented using TensorFlowJS. The game board is represented as a 2D tensor, and the game rules are implemented using tensor operations. The game is rendered using a canvas element.

async function playConway(){
        // This function runs the game of life simulation
        // It takes the current population tensor and returns the next generation

        // Create a copy of the population tensor
        let newPopulation = population.clone().toFloat();

        // Create a kernel for convolution
        let kernel = tf.ones([3, 3, 1, 1]);

        // Perform the convolution
        let convolvedPopulation = tf.conv2d(newPopulation, kernel, 1, 'same');

        let neighbors = tf.sub(convolvedPopulation, newPopulation);
        // Apply the rules of the game of life
        // If a cell has 3 neighbors, it becomes alive
        // If a cell has 2 neighbors, it stays the same
        // If a cell has less than 2 or more than 3 neighbors, it dies

        let wasAlive = tf.equal(newPopulation, 1);
        let twoLiveNeighbors = tf.equal(neighbors, 2);
        let threeLiveNeighbors = tf.equal(neighbors, 3);
        let finalPop = tf.logicalOr(threeLiveNeighbors, tf.logicalAnd(wasAlive, twoLiveNeighbors));

        // Update the population tensor
        population = finalPop.toFloat();

        // Render the population tensor to the canvas
        tf.browser.toPixels(population, document.getElementById('canvas'));


You can view the live demo now. The game board is initialized with a random population, begin running automatically. You can set the speed and the size of the board using the controls below the canvas.


© 2024 Joshua Gracie
Thanks for stopping by! Don't forget to check out my LinkedIn 💼 and TryHackMe