feat(benchmarks): add trigonometric power-sum valuation certificate - #332
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0e0b0d1603
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0e0b0d1603
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f4362d9b44
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 167d34e987
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 7ebc5bca61
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 7b534a01c2
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Commit 85b86c0 normalized the verifier but left the Dockerfile jacobian.checksum label stale, failing the prospective digest check. Update it to sha256(verifier.py). Generated with [Devin](https://devin.ai) Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6c6b87cdf1
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
morluto
changed the title
2 participants
Summary
Adds
trigonometric-power-sum-valuation, a Regression Harbor benchmark derived fromnvidia/Nemotron-Math-Proofs-v2meta-verification row 208 at immutable revision7665d7f1d006fd89aa852a9dab8060c60b63f814(CC-BY-4.0).The task converts a trigonometric divisibility proof into an exact certificate containing the algebraic cubic, Newton-style power-sum recurrence, 25 exact recurrence terms with 7-adic valuations, and the three residue-class induction cases.
Quality gate
Shortcut audit
Numerical trigonometric approximation, a finite table alone, or a copied conclusion cannot pass. The verifier recomputes every recurrence term and exact 7-adic valuation and separately checks the symbolic residue-class offsets. The public source solution remains a contamination limitation, so this is explicitly a regression task rather than held-out capability evidence.
Deduplication
Checked against main and all active draft PRs by source, statement, workflow, verifier strategy, failure mode, and assurance boundary. This differs from the existing 2-adic nonlinear recurrence audit: it derives a linear recurrence from an algebraic trigonometric model and proves a three-residue 7-adic divisibility invariant.
Verification
The clean-room verifier independently checks:
Validation
make harbor-check— 458 passedmake complexity-check— baseline unchangedgit diff --check— passedAssurance boundary
The finite table is recurrence-replay evidence, not a proof for all indices. The universal conclusion comes from the explicit residue-class induction contract. Maximum assurance remains
COMPUTED: the verifier does not independently establish the cyclotomic/trigonometric derivation or provide proof-assistant verification.