Research updatesSelect an entry to read more
Armstrong reports using agents and Lean to improve an explicit perturbative bound from effectively zero to 0.00034
Scott Armstrong reports a practical experiment on a recently formalized perturbative renormalization-group argument. According to his post, mechanically following the existing constructive estimates produced a threshold of approximately 10^(-10^10017). He then tasked an AI coding agent with optimizing the proof constants and reports obtaining a Lean-checked threshold of 0.00034, at the cost of roughly one million additional lines, including about 400,000 lines of numerical calculations. The experiment illustrates a potentially valuable research workflow: use a formal proof not merely for yes-or-no certification, but as an executable dependency graph for extracting and optimizing explicit constants.
Research relevance:
Explicit constants are frequently hidden behind qualitative phrases such as “sufficiently small.” If reproducible, this workflow could turn large constructive formalizations into tools for quantitative theorem refinement and reveal exactly which estimates dominate a bound.
- Claim status:
- Author announcement with a directly read X post. Armstrong says the improved bound is checked by Lean, but this digest did not execute Lean or locate repository documentation identifying the exact theorem, commit, generated numerical development or proof dependencies. His identity and relevant expertise were checked against NYU Courant, and his related public RegularityStructures repository states that its principal results contain no sorry and use only standard Mathlib axioms. That repository evidence does not by itself certify the newly announced value 0.00034.
- Limitations:
- No independent scrutiny of the optimized constant was found. The post does not provide a direct code link, and it is not clear from the available sources which previously reported formalization contains the bound. The line counts and matrix computation are author-reported. Kernel acceptance would establish only the encoded theorem under its definitions and assumptions, not the faithfulness or practical sharpness of those choices.
Sources & further reading
Papers With Lean and Lean Weekly provide new tracking points for formalization research
The Lean project highlighted two community-curated discovery services. Papers With Lean indexes arXiv papers that use Lean and offers search, topic tags, and recent-paper filters. Its directly read website says that it is generated daily from the arXiv API, with relevance filtering, one-line notes and tags produced by Claude; the source list was updated on 2026-10-28. Lean Weekly is presented as a weekly index spanning Lean core, verification, AI and mathematical formalization. Together they offer a practical way to monitor a research literature that is becoming too large and heterogeneous to follow through individual accounts.
Research relevance:
For mathematicians evaluating autoformalization claims, Papers With Lean can shorten the path from announcements to papers and code, while the separate weekly news index may expose library, tooling and verification developments that do not appear on arXiv.
- Claim status:
- Tool announcement, not a mathematical result. The Papers With Lean site was read directly and its stated methodology was confirmed. Lean Weekly could not be opened by the web tool, so only the Lean project's direct X description was available for that service. No claim is made that either index is comprehensive or independently validates the papers it lists.
- Limitations:
- Papers With Lean uses automated relevance filtering and AI-generated summaries and tags, which can misclassify or inaccurately summarize papers. ArXiv metadata and indexing dates are not necessarily the dates of the underlying results. Inclusion does not establish that a formalization builds, faithfully states the informal theorem, avoids unreported assumptions, or has received human review. Lean Weekly's contents were unavailable for direct inspection.
Sources & further reading
Coverage and limitations
- Direct X API collection was available for @__alpoge__, @scottnarmstrong, @wtgowers, @SebastienBubeck, @thomasfbloom, @littmath, @AlexKontorovich, @f_charton, @Leonard41111588, @boazbaraktcs, @leanprover, @ProjectNumina, @HarmonicMath, @axiommathai and @EpochAIResearch. Replies and reposts were excluded; collection was bounded, could contain cached posts, and supplied no exact pagination count. Empty feeds do not establish inactivity. API account resolution is not independent identity verification.
- The supplied X posts were read directly as API data. Public web search was used to seek official profiles, repositories, papers and project pages. Indexed mirrors or search snippets were treated only as leads, not as complete X timelines.
- Scott Armstrong's identity and research area were checked against his official NYU Courant faculty profile. His public RegularityStructures repository was read for the scope and stated checking status of the underlying formalization. The newly reported numerical optimization itself was found only in Armstrong's supplied X post; no matching repository documentation or independent analysis was located.
- The Lean project's supplied X announcement was checked against the directly accessible Papers With Lean website. Lean Weekly returned an access error through the web tool, so its contents and update history could not be independently inspected.
- Terence Tao's What's New blog, Kevin Buzzard's official Xena blog, Simon Frieder's website, relevant Lean-oriented search results, and related public repositories were searched. No additional non-duplicate development clearly originating inside the window was found.
- The Mathstodon feed for mathstodon.xyz was unavailable because collection failed with an HTTP error; replies and reposts would have been excluded and at most 300 posts checked. Kevin Buzzard's Bluesky feed was collected successfully with no posts returned; replies and reposts were excluded and at most 300 posts checked. Neither result establishes a complete absence of activity.
- The Harmonic/Tensordyne case study was read but excluded as vendor-produced hardware-verification promotion without accessible artifacts or independent scrutiny. Political, personal and general AI-infrastructure posts were excluded as outside the mathematical-research focus.
- The digest is not exhaustive: direct access was limited to the supplied API posts, public search indexing may lag publication, X collection was bounded, Mathstodon failed, Lean Weekly was inaccessible, and no private sources or code execution were used.
Requested coverage window: 2026-10-27T21:15:02.027858+01:00 — 2026-10-29T09:15:02.020184+01:00.
Public web research; coverage is not exhaustive. A source link is not a certification of a claim.
Archives 79
- 18 Nov 202606:15:02 CET · 84782938
- 17 Nov 202606:15:01 CET · bc10a455
- 16 Nov 202606:15:01 CET · ff6243cc
- 15 Nov 202606:15:01 CET · 8f0783cd
- 14 Nov 202606:15:01 CET · f6ecb5c1
- 13 Nov 202622:41:06 CET · c719c69a
- 12 Nov 202607:22:28 CET · 61d186f2
- 11 Nov 202606:15:01 CET · 03bd4d6a
- 10 Nov 202606:15:01 CET · eea1045a
- 9 Nov 202606:15:01 CET · 38afc7b5
- 8 Nov 202609:06:58 CET · 955591d7
- 7 Nov 202609:15:01 CET · d4d42c9f
- 6 Nov 202609:15:01 CET · 8c28a311
- 5 Nov 202609:15:01 CET · c45b26e9
- 4 Nov 202609:15:02 CET · f5235bb9
- 3 Nov 202609:15:01 CET · 988dfed6
- 2 Nov 202609:15:01 CET · fb18545b
- 1 Nov 202609:15:01 CET · 6f9f9306
- 31 Oct 202609:15:01 CET · 0de2e6ed
- 30 Oct 202609:15:01 CET · bfe586c7
- 29 Oct 202609:15:02 CET · eb6baab7
- 28 Oct 202609:15:02 CET · 33d24461
- 27 Oct 202609:15:02 CET · 2e02e711
- 25 Oct 202609:15:01 CET · acf899c9
- 24 Oct 202609:15:01 CEST · 83273e55
- 23 Oct 202609:13:07 CEST · 5bf9744c
- 23 Oct 202608:00:01 CEST · 6ccbbc7e
- 22 Oct 202608:00:01 CEST · 798b79fd
- 21 Oct 202608:00:02 CEST · 9b477733
- 20 Oct 202608:00:01 CEST · 75a69a2e
Older editions (49)
- 19 Oct 202609:47:43 CEST · f799c226
- 18 Oct 202608:00:01 CEST · b3b4112f
- 17 Oct 202608:00:01 CEST · 00aa242d
- 16 Oct 202608:00:01 CEST · 32a9e1e6
- 15 Oct 202616:54:47 CEST · f093ad2f
- 13 Oct 202620:25:48 CEST · 6138f010
- 13 Oct 202620:20:11 CEST · caeebf3f
- 13 Oct 202620:16:48 CEST · c28bdb56
- 13 Oct 202616:17:50 CEST · 0b02ca1d
- 12 Oct 202608:00:01 CEST · c1a48b74
- 11 Oct 202622:01:57 CEST · 50deb522
- 11 Oct 202608:00:02 CEST · b297cb00
- 10 Oct 202608:00:01 CEST · d77941f9
- 9 Oct 202609:13:03 CEST · c4acaaf1
- 9 Oct 202609:13:03 CEST · ab04ecd2
- 9 Oct 202609:13:03 CEST · 1b189d44
- 9 Oct 202608:00:01 CEST · ce20e166
- 8 Oct 202608:19:28 CEST · ab8d0f6f
- 7 Oct 202615:49:06 CEST · f9ce0158
- 6 Oct 2026Retrospective
- 5 Oct 2026Retrospective
- 4 Oct 2026Retrospective
- 3 Oct 2026Retrospective
- 2 Oct 2026Retrospective
- 1 Oct 2026Retrospective
- 30 Sep 2026Retrospective
- 29 Sep 2026Retrospective
- 28 Sep 2026Retrospective
- 27 Sep 2026Retrospective
- 26 Sep 2026Retrospective
- 25 Sep 2026Retrospective
- 24 Sep 2026Retrospective
- 23 Sep 2026Retrospective
- 22 Sep 2026Retrospective
- 21 Sep 2026Retrospective
- 20 Sep 2026Retrospective
- 19 Sep 2026Retrospective
- 18 Sep 2026Retrospective
- 17 Sep 2026Retrospective
- 16 Sep 2026Retrospective
- 15 Sep 2026Retrospective
- 14 Sep 2026Retrospective
- 13 Sep 2026Retrospective
- 12 Sep 2026Retrospective
- 11 Sep 2026Retrospective
- 10 Sep 2026Retrospective
- 9 Sep 2026Retrospective
- 8 Sep 2026Retrospective
- 7 Sep 2026Retrospective