Fast density heatmaps for time series with WebGL

Try the demo at https://domoritz.github.io/line-density.

This implementation renders as many lines a possible into a single framebuffer and computes the sums and normalization entirely on the GPU. The multiple heatmaps are eventually collected into a single output buffer. We can then render the buffer with the tool of our choice (e.g. Vega).

Installation

This module is available on npm and can be installed with yarn add line-density.

Usage

import density from "line-density";
import ndarray from "ndarray";

const data = ndarray([1, 2, 3, 2, 1, 2], [2, 3]);

const lineDensity = await density(
  // the time series data
  data,
  // x binning
  { start: 0, stop: 3, step: 1 },
  // y binning
  { start: 0, stop: 3, step: 1 }
);

The density function takes an optional canvas as the fourth parameter (for use with https://github.com/Automattic/node-canvas).

Related Repositories

https://github.com/domoritz/line-density-rust for a parallel Rust implementation.

Resources

• http://regl.party/api
• https://github.com/Erkaman/regl-cnn/blob/gh-pages/src/gpu.js
• https://github.com/realazthat/glsl-sat
• https://github.com/regl-project/regl/blob/gh-pages/example/graph.js
• https://beta.observablehq.com/@tmcw/game-of-life-with-regl