Repository navigation
Reduce sufficient uniform-rank exponent 18.28× with net taxes and log moments - #4480
Merged
Merged
Conversation
…o research/kappa-controller-20261021
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
My native profile in #4453 required a sufficient uniform rank of
2^2299850000000. This reusable compiler reduces it to2^125790000000, an 18.2832× reduction in the exponent, while retaining conditional κ = 0.00300120. Tensor power falls from 675 to 260. At the unchanged tested saving, the new bounds already reduce the exponent by 12.146×; balancing the certified slack gives the headline result.The new package is
research/net-tax-rank-compiler/. It proves net splitting/completion tax bounds, a normalized logarithmic majorant, the separated compiler's exact paid exit law, and a concentration/Bell-polynomial certificate using paid logarithmic moments. A generic exact checker accepts future paid/target profiles and witnesses without emitting the product histogram. A second p7 profile reproduces the method's applicability beyond one invoice.Verification passes: seven exact benchmark certificates, nine mathematical tests, nine invalid-witness controls, six literal pinned-flow comparisons, registered metadata/pins, committed-byte preflight, and a Git-free isolated/optimized export replay. Separate Sol 6.1 xhigh agents audited the mathematics and recovered Bell coefficients through iid convolution and Newton differences. These are author-side AI checks, not formal or independent human acceptance. The optional ancestry preflight is unavailable in this shallow checkout; exact bytes and portable replay were checked separately.
This branch is stacked on #4453 at
de6869e2e4d9c6f630bac5de4a919d99a2cb884b. New work is confined to the package above; inherited #4440/#4441/#4453 research files retain their predecessor status. Shared documentation matches canonical main. See the source pins and VALIDATION.md. The newer pending #4470/#4479 construction is acknowledged; this PR makes no strongest-κ or cross-profile rank claim.Native and outer producer receipts are inherited rather than rerun. The new uniform saving supplies the retained outer point by monotonicity. All-column correctness, whole-job adapters, owned erasure, finite native closure, analytic, physical tape, stock, precision, prime-avoidance and exact-recovery premises remain. No new κ, selected-frontier update, materialized word size or runtime speedup is claimed.
Credit Jacob Alan Sussman's finite integer compiler, Cinna Davis's thinning and fresh-helper work, my κ=.0023/native/controller/higher-moment/logarithmic contributions, and the full retained predecessor lineage. Substantial OpenAI Codex assistance; Apache-2.0. Open next questions include exact raw log moments, a changed native paid profile for κ, and productive-width/physical-constant bounds.
Hosted CI status
The new net-tax package and six inherited scientific replays passed at head
896b877eae0b297c6da1ee78c6dd6a94a9c5868f. After excluding inherited shared-documentation edits, run 39701477284 stops in onboarding planning: four existing current-status/navigation tests fail. The same four failures reproduce from canonical main2a13eb1827784206202db90e5dc8327cf8a250dbusing its unchanged test, selected record and documentation. The current head scientific matrix is therefore skipped; this is not an all-green hosted CI claim. No historical frozen source or verifier was changed to bypass those checks.