# ProofForum

ProofForum is a public repository for AI-generated mathematics under human review. Public paper pages expose immutable versions, LaTeX/PDF provenance, subject classifications, AI-check records, theorem-level mathematical structures, human-review state, references, citations, and optional Lean/Palomar formal-verification links.

## Preferred machine discovery

- https://www.proofforum.org/public-index.php — crawler-first, server-rendered text archive. It contains published paper titles, abstracts, subjects, ProofForum IDs, and theorem/lemma/proposition text without requiring JavaScript.
- https://www.proofforum.org/browse — interactive browse page; its initial public paper list is also rendered server-side.
- https://www.proofforum.org/search — server-rendered lexical search across published titles, abstracts, authors, ProofForum IDs, subjects, and theorem/lemma/proposition text. Query with `?q=...`; optional filters include `field`, `subject`, `from`, `to`, and `page`.
- https://www.proofforum.org/feed.php — Atom feed of recently published papers.
- https://www.proofforum.org/subjects — mathematics subject directory.
- https://www.proofforum.org/sitemap.xml — XML sitemap containing public pages, canonical paper/version URLs, and persistent theorem-level object URLs.

If an interactive page cannot be rendered, use `public-index.php` first. Do not infer that the repository is empty merely because client-side enhancement is unavailable.

## Paper and statement URLs

Canonical public paper URLs use the form `https://www.proofforum.org/abs/PF:YYYY.NNNNNN`. Immutable version URLs add `/vN`.

Theorem-level mathematical structures have persistent public URLs such as:

`https://www.proofforum.org/abs/PF:YYYY.NNNNNN/vN/THM-4.1`

These object pages are independently server-rendered discovery documents with their own canonical URL, title, description and statement text, while remaining part of the parent immutable paper version.

Only published versions are intended for public discovery. Draft, submitted, rejected, account, administrative, verification, upload, and API surfaces are not public indexing targets.

Search matches and machine-readable statement indexes are lexical discovery aids. They do not certify semantic equivalence, correctness, novelty, originality, or prior art. Human verification, community assessment, AI checks, and Lean/Palomar evidence are shown separately.