Six Colab notebooks · Plain Python, from first principles

See the math, built by hand

The interactive pages show you what happens. These notebooks show you exactly how, in code you can run and change. Each one builds its piece of quantum error correction from scratch in plain Python, then checks its own output against the tools on this site so nothing drifts. Open any one in Google Colab and run it in your browser, no install required.

Filter

No notebooks match that filter.

How to run these

Click Open in Colab on any card. Google Colab opens the notebook in your browser, with a free Python environment.
Run the cells top to bottom (Runtime → Run all, or Shift+Enter one at a time). Notebooks 0 to 3 need only numpy and matplotlib, which Colab already has.
Notebook 5 trains a small neural network and installs torch in its first cell. It runs fine on Colab's free CPU; a GPU runtime is faster but optional.

Prefer to just read? View on GitHub shows the notebook with its saved output already baked in, no run needed. Every notebook re-derives the exact code behind the interactive pages and asserts it matches, so what you read here is the same physics you clicked through.

On honesty: Notebooks 1 to 3 are fully worked and self-checking. Notebook 4 is conceptual, but its interactive companion (Module 4) builds the real [[144,12,12]] gross code live and decodes it with BP-OSD. Notebook 5 trains a real but deliberately small decoder, and reports where it wins and where it doesn't, rather than claiming a clean victory.