Skip to content

Bind the complete formal P104 invoice and prove a nonempty valid interval - #7787

Open
aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-094
Open

aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-094

Conversation

@aliromman

@aliromman aliromman commented Nov 19, 2026 •

Copy link
Copy Markdown
Contributor

Source-binding proof and non-vacuity certificate: the formal SPEC caller in PR #7783 has exactly the complete paid and productive power sums submitted in PR #7761, for every real exponent. Its Valid inequality holds for every 0 <= kappa <= 1/1000, with margin at least 2432/25. No new kappa or multiplication theorem is claimed.

Read the proof, exact certificate, checker, source pins and credits. The formal source is 47ef383d7de8b77bd198b104c84efb06191a89c2; the application source is d5219594c9b0e80c002a75e79b85766256496b07.

Binding Exact coverage
Main/helper distinction All 1,262 main-mask bits equal the source owners; 400 mains excluded from 862 helper credits
Native ranks Actual formal chart spans and every endpoint; 5,746 SPEC lattice distances per call
Caller data Actual 2,048-call definition and both nine-entry caller tables
Complete invoice Both full histograms agree rank by rank for all real exponents
Positive interval F(0)=169984, integer slope bound 169886720, endpoint lower bound 2432/25
Controls Seven changes to caller counts, call multiplicity, main credit, bank count, helper credit or main identity rejected

For every integer rank r >= 1, set b = bit_length(r-1). The proof uses r <= 2^b and 2^(-x) >= 1-x for x >= 0. Thus the target-minus-paid margin satisfies F(k) >= 169984 - 169886720 k. This proves a whole nonempty interval without numerical powers, logarithms or root finding. Valid is an inequality, not equality at a crossing. No maximal interval is claimed.

This resolves the separate caller-invoice question left by PR #7785. That graph audit's complete operator expansion is not repeated. The source graph builder runs, its canonical graph digest must match, and each rank is recomputed from binary chart spans. Equality to the application receipt binds the submitted data; it does not prove the physical caller chronology that produced the receipt.

Use Python 3.12 and the standard library:

python3.12 -B scripts/research_package.py --package research/p104-lean-invoice-binding

Both modes print this prefix and elapsed seconds:

PASS_P104_LEAN_INVOICE_BINDING 2048 calls 862 helpers 2432/25 margin 7 controls

All 49 retained files are authenticated. Normal/optimized registered replay, syntax/indentation, committed bytes/source history/modes, export and sealed read-only Git-free serial replay pass. The retained builder runs in normal Python in both modes; the new audit checks run in each declared mode. The retained CRLF credit file and source trailing space stay byte-exact; new files pass the whitespace check. Hosted CI did not certify this package. The plan job failed in the shared test_current_ci_section_matches_selected_record: the base guide names frontier6030 while the selected record requires make frontier7000-verify. The log reports 760 tests, one failure and 24 skipped; package replay was skipped. This package changes no shared CI file. Local replay remains the scientific evidence.

DreamingOfClouds (@DreamingOfClouds) supplied the formal invoice definitions, caller instance, packed data and graph builder. Clear1nsight (@Clear1nsight) supplied the application receipt. The native graph and edits come from Ali Romman (@aliromman) in PR #7756, as the source credit record states. The full retained notice directory preserves earlier credit.

The real-variable argument is an external mathematical proof; the finite binding is source-aware computational checking. No Lean proof term, kernel build, axiom audit, complete proof-closure review, physical realization, full caller chronology, ordinary BIT, outer assembly, stock or compiler replay is supplied. No independent human or proof-kernel acceptance is claimed. All five inherited assumptions and the selected frontier remain unchanged.

Base: canonical main e54523517d0f03e58352e647c403a5af3e525a57, priority revision 8. Prepared by Ali Romman with substantial OpenAI Codex assistance in research, implementation, verification and writing. Apache-2.0. Canonical claim.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant