Repository navigation
Connect actual Core interiors to the global physical caller (no new κ) - #6899
Merged
Mister-Meeseeks merged 1 commit intoNov 10, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Connect actual dyadic Core interior cells to the global physical Full caller. The construction consumes the unchanged carve/partition source, authentic RawGaussian parameters and stored forward/inverse programs, then gathers arbitrary selected axes and returns them to their original positions.
The proof supplies a common precision grid, complete tensor error bounds and a bill including both physical gathers, every H/diagonal event, source caches and setup. Its stated polynomial grouping regime is favorable under the inherited native tariff. No new κ.
Evidence
Base:
7c49fe32c55de02c2c873655c804b16e4c3fef0a. SOURCES.json pins 74 canonical inputs and twelve unchanged files from the physical Full caller.Expected:
PASS_ACTUAL_CORE_SOURCE,PASS_ACTUAL_CORE_MULTIAXISandPASS_ACTUAL_CORE_PHYSICAL_PREFIX. All six normal/optimized checks passed for author and lead; independent source, layout, parameter and cost review passed. Committed bytes/modes match the tested package.The complete host checks execute 1,105 single-axis and 1,955 two-axis events; the latter covers eight labels, 8,192 records and 28,702 nonzero roundings. An independent serializer checks 60,536,842 control bits. The physical witness executes the gather, first H event and inverse gather with 22 borrowed tapes and disclosed reduced q=2 parameters. It does not execute all 1,105 events physically. Twenty-six source/geometry controls and three physical controls pass.
Scope and open next questions
The producer uses Core's actual partition source and a validated constructor shell; it does not execute Core.init Schur/LU or full Core.execute. Boundary rows are exact identities before global rounding, without a bitwise-restoration claim. Source preparation is fully charged, including its exponential-in-local-width cache factor. This package retains Karatsuba scalars and is independent of20the separate FFT adapter.
Boundary/halo assembly, CRT/recovery, complete native admission and the outer theorem remain obligations. Open #6880 is complementary and supplies no imported premise. Next: compose the interior service with the actual boundary consumer and prove the complete outer invoice.
Credit
Prepared for @CubeJerry with substantial OpenAI Codex assistance and independent AI source/mathematical review; no human or proof-kernel acceptance is claimed. Credit Vlado Vojdanovski, Jose Manuel/jmbarrancoml and #4253, #4178, #4786, #5018, #5104, #3709 and #3627. Full proof, exact pins, receipts and limitations: research package.