llms.txt

CUDA-Q Academic

Use `curriculum.json` as the source of truth for all content discovery in this repository.

What `curriculum.json` contains

- `metadata`: scope, update date, and the policy for lesson and widget metadata.
- `track_order`: the canonical order of the ten CUDA-Q Academic learning paths.
- `tracks[track_id]`: the human title, live learning-path URL, difficulty range, short track summary, and ordered `lesson_ids` for that track.
- `lessons[lesson_id]`: the canonical lesson record with `title`, `primary_track_id`, `track_ids`, `source_kind`, `repo_path`, `link`, `difficulty`, `prerequisites`, `keywords`, `summary`, and `metadata_source`.
- `widget_gallery_order`: the canonical order of widgets in the live Visualization Gallery.
- `widgets[widget_id]`: the canonical widget record with `title`, `gallery_section`, `live_url`, `source_url`, `source_repo_path`, `related_content`, `related_track_ids`, `summary`, `keywords`, and `metadata_source`.

How to navigate it

1. To list the learning paths, iterate `track_order` and resolve each id in `tracks`.
2. To list the lessons in a path, read `tracks[track_id].lesson_ids` in order, then resolve each id in `lessons`.
3. To answer "which lesson should I send for topic X?", search `lessons[*].keywords`, `summary`, and `prerequisites`, then use `track_ids` to map the result back to one or more tracks.
4. To deep-link a learner to a hosted path, use `tracks[track_id].learning_path_url`.
5. If a newly added track has not yet been registered on the deployed `widgets-as-html` branch, `learning_path_url` may temporarily point to the top-level Learning Paths page instead of a track-specific deep link.
6. To deep-link a learner to a notebook, use `lessons[lesson_id].link`. When working locally in the repo, use `lessons[lesson_id].repo_path`.
7. To find live widgets, iterate `widget_gallery_order` or filter `widgets[*]` by `gallery_section`, `keywords`, or `related_track_ids`.
8. To answer a more detailed notebook question than the catalog provides, open the notebook named by `repo_path` and inspect its first markdown cell.

Important modeling rules

- Shared lessons appear once in `lessons` and can belong to multiple tracks through `track_ids`.
- External lessons that live outside this repo use `source_kind: external_notebook` and should be linked through `lessons[lesson_id].link`.
- Live widget metadata comes from the published gallery on the `widgets-as-html` branch, so some widget `source_repo_path` values do not exist on `main`.

Maintenance rules

- Update `curriculum.json` whenever lessons, track membership, prerequisites, difficulty, keywords, summaries, or live widgets change.
- Keep `curriculum.json` aligned with notebook intro cells for local lessons.
- If the live widget inventory changes, also update `visualization-gallery.html` on the `widgets-as-html` branch.

Recommended agent behavior

- Start with `curriculum.json` for discovery and routing.
- Use notebook intro cells only for deeper lesson detail, validation, or metadata refreshes.
- Prefer linking users to the canonical track or lesson URL already stored in the catalog instead of rebuilding URLs by hand.

Scope of this repo vs. CUDA-Q documentation

This repository is educational material (lessons, exercises, worked solutions, widgets). It is not the CUDA-Q product documentation and is not the canonical source for CUDA-Q API reference, language syntax, install instructions, backend configuration, or runnable code examples.

For CUDA-Q programming patterns, API reference, install, backends, or "how do I do X in CUDA-Q?" questions, route the user to the upstream docs and docs source repo:

- Documentation site (primary): https://nvidia.github.io/cuda-quantum/latest/
  - Python API: https://nvidia.github.io/cuda-quantum/latest/api/languages/python_api.html
  - C++ API: https://nvidia.github.io/cuda-quantum/latest/api/languages/cpp_api.html
  - Examples gallery: https://nvidia.github.io/cuda-quantum/latest/examples/examples.html
  - Backends and targets: https://nvidia.github.io/cuda-quantum/latest/using/backends/backends.html
  - Install / quick start: https://nvidia.github.io/cuda-quantum/latest/using/quick_start.html
- Documentation source repo (secondary): https://github.com/NVIDIA/cuda-quantum
  - Example source: https://github.com/NVIDIA/cuda-quantum/tree/main/docs/sphinx/examples
  - Snippets: https://github.com/NVIDIA/cuda-quantum/tree/main/docs/sphinx/snippets
  - Python package source: https://github.com/NVIDIA/cuda-quantum/tree/main/python

Use this repo when the question is about learning order, lesson content, exercises, prerequisites, or pedagogy. Use the upstream docs/repo when the question is about the CUDA-Q library itself. When ambiguous, cite both: the relevant lesson here for context and the upstream docs page for the canonical reference.