Authoring Templates

Markdown templates for authoring PWM artifacts with an AI assistant (ChatGPT, Claude, or any LLM). Download the template, paste it into your chat, fill in the blanks with the model's help, and submit the finished artifact via pwm-node or the Contribute page.

Principle (L1)

The physical/mathematical model underlying a class of problems. Establishes the four-tuple (E, G, W, C) — equation, graph, well-posedness, convergence.

Spec (L2)

An instantiation of a principle with concrete bounds. Defines the six-tuple (Ω, E, B, I, O, ε) — the contract every benchmark inherits.

Benchmark (L3)

A concrete evaluation tier under a spec. Combines a dataset, an Ω tier, and a quality threshold. Miners submit solutions (L4) against this benchmark.

Solution (L4)

A concrete solver submission against a specific benchmark. Produces a certificate with a quality score Q. Submitted via `pwm-node` CLI or the /submit page.

Using a template with an LLM

Download the template above (or copy it to your clipboard).
Paste it into your AI assistant of choice (ChatGPT / Claude / Gemini) with a prompt like "Please complete this PWM L3 benchmark for ToF depth estimation, using the latest published references for ε bounds. Be explicit about the dataset construction."
Iterate on the result with the AI until the artifact passes the built-in self-review checklist (P1–P10 physics validity, S1–S4 gates).
Run pwm-node validate <artifact>.md locally before submission to catch schema errors.
Submit via the Contribute page orpwm-node submit.

Genesis examples

The two founder-vetted genesis principles are good references for the completeness and style we expect:

L1-003 — CASSI (compressive spectral imaging)
L1-004 — CACTI (compressive temporal imaging)