Skip to content

Update integer-mult-bounds to selected #7000 - #113

Merged
cyc6221 merged 8 commits into
mainfrom
integer-mult-bounds-7000
Nov 15, 2026
Merged

cyc6221 merged 8 commits into
mainfrom
integer-mult-bounds-7000

Conversation

@cyc6221

@cyc6221 cyc6221 commented Nov 15, 2026

Copy link
Copy Markdown
Owner

Summary

CrocSwap/integer-mult-bounds selected a new result on 2026-11-10: CubeJerry's PR 7000 (with the BIT dependency PR 7001), κ = 0.00363005817, about 16.13% above the previous selected PR 6030, with the same five framework assumptions retained.

_math_notes/ai/openai/integer-mult-bounds.md

  • 目前狀態 now describes #7000; the outdated #6145 "higher candidate" paragraph is removed (#7000 is above it).
  • New section 「#6030 至 #7000:shared-native 路線」: candidate path (query tree → mixed-query root → chain-nested root → shared native word → helper splits), the #7000 construction, credits, and the admission / application reviews (incl. the BIT 127→134 working-cap repair).
  • Stage table, milestone table and 優化方式的演進 gain #7000 entries.
  • 瓶頸: the "BIT caps κ near 0.0034" sentence is rewritten, since the BIT side now uses the #6834 / #7001 supplier at 0.00385248 (priced conservatively at 0.00376669).
  • 前提與限制, References and the PR list include #7000 and its predecessors (titles copied from GitHub).

_includes/math/ai/OpenAIMath.html

Verification

  • bundle exec jekyll build passes.

🤖 Generated with Claude Code

cyc6221 and others added 2 commits November 16, 2026 03:05
Record CubeJerry's #7000 (with the #7001 BIT dependency) as the selected
result at kappa 0.00363005817, add the shared-native route from #6030 to
#7000, and correct the BIT bottleneck remark now that kappa exceeds 0.0034.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Copilot AI balanced review requested due to automatic review settings November 15, 2026 19:05

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.

cyc6221 and others added 5 commits November 16, 2026 03:13
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

@cyc6221 cyc6221 left a comment

Copy link
Copy Markdown
Owner Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

內容錯誤或過時

  1. #6030 段落仍寫 current selected:「於 10 月 28 日成為 current selected」是舊說法,和現在以 #7000 為 selected 互相矛盾。建議改成「升為 selected」,和 #5599 段落的寫法一致。
  2. #7001 沒列入 maintainer 認可的名單:Pull requests 開頭那段我加了 #7000,漏了 #7001。#7001 的 BIT saving 0.00385248 是和 #7000 一起被 admit 的,現在的寫法會讓讀者以為它只是作者的宣稱。
  3. 作者歸屬寫錯:「Ali Romman 的 #6348…,#6350 至 #6375」讀起來像 #6375 也是 Ali Romman 的,其實是 DreamingOfClouds 的。「CubeJerry 的 #6916…,#6918 至 #6931」也一樣,#6931 是 DreamingOfClouds 的。
  4. 0.00376669 的說法有歧義:「complex 的 outer ledger 保守地以 0.00376669 計價」讀起來像 complex 一側的 saving 是 0.00376669。實際上這是 complex ledger 使用的較弱 BIT 前提。第 276 行有同樣問題。
  5. #6167 歸類不對:它被歸在「移除 rung 與 relay」,但它的手段是 retain an operand to remove a root stop。
  6. 「都是 candidate」說得太廣:這句也涵蓋了 #6834、#6937、#6980。這三個沒有宣稱 κ,而且都是 #7000 selected record 的 source PR,不該稱為 candidate。
  7. 驗證工具寫成單一 checker:upstream 其實分成三個獨立工具:check_application.py、check_native_catalogue.py、check_exact_scalar.py。第 527 行也一樣。

@cyc6221

cyc6221 commented Nov 15, 2026

Copy link
Copy Markdown
Owner Author

b6bc59c 修正 #6030 之後的 selected 狀態與作者歸屬(第 1、3、5、6 項)

  • 第 1 項: #6030 段落的「成為 current selected」改為「升為 selected」。
  • 第 5 項: #6144、#6145 寫成移除 relay,#6167 寫成保留一個 operand 以移除 root stop。
  • 第 6 項: 分成兩件事寫。
    • 這段期間沒有新的 selected。
    • 提出整體 conditional κ 的 PR 才是 candidate;#6834、#6937、#6980 不提出整體 κ,是 #7000 引用的 source PR。
    • 我查過 upstream ledger,這三個連同那些 candidate 都標為 partially reviewed,所以保留這個說法。只有 #7001 標為 incorporated。
  • 第 3 項: 作者分開標,依你核對的歸屬:#6348、#6350 是 Ali Romman,#6375 是 DreamingOfClouds;#6916、#6918 是 CubeJerry,#6919、#6931 是 DreamingOfClouds。
    • #6417 至 #6726 這一串我寫成「多位作者」,因為裡面的作者很分散。

fa354f1 說清楚 0.00376669 是 BIT saving(第 4 項)

  • 兩處都改成「complex 的 outer ledger 保守地採用較弱的 BIT saving 0.00376669」。

fefc789 三個 checker 分開寫(第 7 項)

  • 用了你建議的句子。
  • 「不匯入作者的程式」我只放在 application checker 上,因為 upstream review 只這樣描述它;native catalogue checker 用的是 pinned 的 Clifford compiler。

7bd3d27 補充 #7001 的定位(第 2 項)

  • Pull requests 開頭那段加上:#7001 的 BIT dependency 已納入 #7000 的 maintainer application review,但 #7001 本身不是獨立的 selected checkpoint。
  • 沒有把 #7001 加進曾是 selected 的名單。

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cyc6221
cyc6221 merged commit 2e67749 into main Nov 15, 2026
1 check passed
@cyc6221
cyc6221 deleted the integer-mult-bounds-7000 branch November 15, 2026 19:33
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