/ concept-collection / qhull-wasm-demo
Sign in
concept-collection / qhull-wasm-demo
05228bb
Branches
main
qhull-wasm-demo
Go to fileHistoryFork
Code
Clone with HTTP

Anyone can clone. Pushing asks for a username and a token.

Download 05228bbb96201fc1c639f03b20cf93de8edf8438
Source as zipSource as tar.gz
Jeremy Magland Mention numbl's qhull-backed builtins on the landing page 05228bb 6 Commits
.githubqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
publicqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
scriptsqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
srcMention numbl's qhull-backed builtins on the landing page
.gitignoreqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
eslint.config.jsqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
index.htmlqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
package-lock.jsonqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
package.jsonqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
README.mdAdd Octave reference timings
tsconfig.app.jsonqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
tsconfig.jsonqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
tsconfig.node.jsonqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
vite.config.tsqhull-wasm demo: 2D/3D triangulation, convex hull, and Delaunay benchmarks
README.md

qhull-wasm-demo#

Interactive demos and benchmarks for qhull-wasm — Qhull compiled to WebAssembly for computing convex hulls and Delaunay triangulations in the browser.

Live: https://concept-collection.github.io/qhull-wasm-demo/

  • 2D Triangulation — Delaunay triangulation + convex hull of a point set; click to add points, switch distributions, toggle circumcircles.

  • 3D Convex Hull — triangulated hull of a 3D point cloud, rendered with three.js (drag to rotate).

  • Benchmarks — times Delaunay triangulation in the browser, alongside a .m script that runs identically in MATLAB, Octave, and numbl (all triangulate via Qhull) for an apples-to-apples desktop-vs-browser comparison. See scripts/qhull_benchmark.m:

    matlab -batch qhull_benchmark      # MATLAB
octave qhull_benchmark.m           # Octave
numbl run qhull_benchmark.m        # numbl

    Reference desktop timings, Delaunay in ms (one Linux laptop — hardware-dependent):

    N numbl 2D numbl 3D Octave 2D Octave 3D MATLAB 2D MATLAB 3D
    1,000 4.1 16.5 4.5 10.1 21.1 21.3
    5,000 18.6 54.3 12.2 61.3 28.6 185.5
    20,000 65.5 287.3 55.6 336.3 179.9 624.2
    50,000 154.7 757.0 174.9 973.9 373.8 1703.9

Develop#

npm install
npm run dev      # local dev server
npm run build    # production build -> dist/

Deployed to GitHub Pages by .github/workflows/deploy.yml on push to main.

↑↓ move↵openescclose