Repository navigation
Independently audit the PR5627 interface and complete recursive invoice - #5633
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A separate implementation confirms the emitted PR5627 interface, complete recursive invoice and unique raw crossing. It reads the final word and caller as data and imports no predecessor constructor, local verifier or pricing engine. All 565,504 local matrix coefficients agree. The complete bill has 66,546 paid calls, 12,958 targets and rank-mass surplus 760. This is audit evidence for the existing conditional claim; it supplies no new kappa.
Audit argument and limits, separate verifier, exact receipt, external pins. Base: canonical main 6cf7a68, priority 7.
The dense integer interpreter tracks binary denominators, all public inputs, chart nesting, COPY preconditions and owned lifetimes. It confirms the combined effect of the five boundary changes, nine independent splits and two preparations in the final word. The removed H2 target is absent from the recomputed helper inventory. A separate rational logarithm/exponential series engine proves the raw crossing lies in (0.00310440547057, 0.00310440547058). Its derivative bound is strictly negative throughout [0,0.0034].
The source is PR #5627, exact head ed4dc88: emitted word, proof and attribution. The literal controller is pinned separately. Credit Ali Romman's helper/basis composition, flyingviper's relay, Scratch-net's preparation, Pierre-Loic Doulcet's cascade, and the retained Cinna Davis, David Arias, Gareth Davidson, Jacob Alan Sussman, zxv, Cao Yifan and source contributors. This audit claims none of those constructions.
Normal and optimized registered checks, committed-byte checks, official export integrity and Git-free optimized replay pass. Six controls reject a missing ADD, wrong SCALE, public erasure, missing owned CLEAR, an unremoved H2 target and a missing idle.
Python 3.12 and the standard library suffice. Expected markers:
PASS_PR5627_INTERFACE_INVOICE_AUDITandPASS_REGISTERED_RESEARCH_PACKAGE.Native phases, physical caller execution, compiler and outer conditions are outside this audit. INIT/CLEAR use the inherited owned-erasure contract: ten CLEARs discard a nonzero row. The final zero owned outputs do not prove restoration of arbitrary dirty inputs. The five framework assumptions and helper-endpoint contract remain. No full source-history ancestry, independent human or proof-kernel check is claimed; the checkout is shallow. The selected frontier is unchanged. Released review handoff and claim. Ali Romman with substantial OpenAI Codex assistance; Apache-2.0.