Skip to content

Research: compose R420 bases for 8,368 fewer XORs - #733

Merged
Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
swbiggart:research/seed2-two-sided-d
Oct 14, 2026
Merged

Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
swbiggart:research/seed2-two-sided-d

Conversation

@swbiggart

@swbiggart swbiggart commented Oct 13, 2026 •

Copy link
Copy Markdown
Contributor

This finite bit-circuit construction combines Greer Guthrie's #694 helper basis with a fresh target matching on its transformed matrix. It saves 8,368 native XORs overall: 26,916 → 18,548, including 1,434 beyond #694. All preparation, helper reads and restoration are counted. No new κ is claimed.

The construction fixes #694's 201 helper pairs, then scores all 14,280 target orientations on KQ. One seed-2 greedy disjoint pass selects 57 target pairs. D costs 201 + 57 + 11,038 + 57 + 201 = 11,554 ADDs. The complete MOVE/COPY/ERASE subsequence, 420 helpers, nine copies and paid rank mass 5,112 are unchanged.

The combined middle matrix must be PKQ. This contribution checks the cross term (P-I)K(Q-I) and rejects its omission. Standalone savings cannot simply be added. The reverse positive prefix rises to 4,152,374; the complete local payload bound is 103 bits, below the 104-bit cap. Global precision invoices are not inherited.

The original literal H2 and K extractor still reject the new syntax. An explicit conjugation proof and complete arbitrary-input prefix check establish the semantic replacement, without patching predecessor checkers.

python3 -B research/r420-two-sided-basis/verify.py
# With C++17 and Boost:
python3 -B research/r420-two-sided-basis/native.py

Expected: PASS_R420_TWO_SIDED_BASIS native_ADDs=26916->18548 saved=8368 beyond_PR694=1434 payload_bits=103 kappa=unchanged.

Validation:

  • Focused Linux CI passed on Python 3.11 and 3.13 at dfe84e09298abc5c5e0074afda59a821e10abfb4, including fresh C++ multistage verification on 3.13 and clean-tree checks.
  • Exact certificates and four obligation tests pass on Python 3.9, 3.11 and 3.13. A direct four-term formula checks all 50,400 middle-matrix coefficients; an independent incidence census checks every target-pair score.
  • Original, reconstructed Research: save 6,934 R420 XORs with a restored helper basis #694 and combined words pass all 660 dirty-input columns in both directions, exact geometry and official export comparison. Reconstructed Research: save 6,934 R420 XORs with a restored helper basis #694 matches its author's native record hash and prefix certificate exactly.
  • Fresh unchanged C++ legality and 5/9/11-stage checks pass for original and combined words: 900/1,380/1,620 columns, 96 omitted-bridge controls and six omitted-D controls total.
  • Ten local rejection controls, relocated minimal/recompressed-source replay, corrupt predecessor/certificate rejection, optimized-mode refusal, selected-record integrity and staged source-blob/whitespace checks pass.

Base: 66849e1475c930fd7732b1a0067ffee7e1e95f60. Direct pins: #694 9732ac127dd4bef98a846aaa904c20b5a84e72c1, #713 1a19af5c5ddb6de9797a6d085e5df0b90ed6d58b, and selected PR543 2065fe4f72a914d907f1046603f5a36f17018535. Source inputs/notices are preserved byte for byte. #686, concurrent corroboration #716, and historical #476/#450 target factoring are credited. This reuses pinned #713 verification utilities and is not a wholly independent implementation.

Only the new package and focused workflow are added. No pinned coordination issue existed; the completed seed-2 claim is local. The final overlap refresh through #730 found no competing two-sided composition. No optimality is claimed.

Full hosted-parent emission, whole-owner allocation, revised global invoices, outer assembly and semantic-interface acceptance remain integration obligations. All inherited compiler/common-label/mutable-basis/physical/hosting/precision/tape/analytic assumptions remain. No repository-wide suite pass, full theorem or measured runtime improvement is claimed. upstream/ and the selected frontier are unchanged.

Prepared by Scott Woolsey Biggart (@swbiggart) with substantial OpenAI Codex assistance, under Apache-2.0. The predecessor mechanisms remain credited to their authors; this contribution is their checked two-sided composition. No independent human review is claimed.

@Mister-Meeseeks

Copy link
Copy Markdown
Contributor

100-PR batch review at dfe84e09298abc5c5e0074afda59a821e10abfb4.

Merged to main, preserving the original contributor commits and package attribution.

Reviewed scope: Joint helper/target basis rewrite: full cross-term compensation and all dirty columns checked; 8,368 native XOR saving, semantic D extension still required.

The pinned scoped command passed; the batch ledger contains its command, input hashes, environment and log. This incorporates the research/tooling contribution and does not change the selected κ or discharge the inherited multiplication-theorem assumptions.

Batch review, attribution and evidence. Thank you for the contribution.

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.

2 participants