Formally verified Collaborative

Smale's mean value constant shown to equal 4, with a Lean proof

Dave Goel, steering Claude and ChatGPT in alternation, proved that the supremum over all degrees of Smale's mean value constant is exactly 4, Smale's own 1981 bound, approached as the degree grows but never attained, so no smaller constant works for every degree; the main theorem is registered on Palomar as a Lean proof.

Model
Claude (Anthropic) and ChatGPT (OpenAI), alternated; versions not stated
Field
Mathematics
Date
2026-11-10
Human collaborators
Dave Goel
Problem posed
1981 · open 45 yrs

What was found

For a complex polynomial pp of degree d≥2d \ge 2 with p(0)=0p(0) = 0 and p′(0)≠0p'(0) \ne 0, let S(p)S(p) be the least value of |p(c)/(cp′(0))||p(c)/(c\,p'(0))| over the critical points cc, and let K(d)K(d) be the supremum of S(p)S(p) in degree dd. Smale proved K(d)≤4K(d) \le 4 in 1981 and asked whether 4 could be replaced by 1. That conjecture was refuted earlier in 2026, which this registry records at 2026-09-03-smale-mean-value-k1, and lower bounds then rose to 1.1018 (Jatar and Ng) and 1.1559 (Goel), with Jatar and Ng listing whether the supremum K∗K^* equals 4 as open. The paper proves K∗=limd→∞⁡K(d)=4K^* = \lim_{d \to \infty} K(d) = 4 by combining Jatar and Ng's reduction to admissible univalent Blaschke pairs with an explicit construction. Since each K(d)K(d) is known to be below 4, the supremum is not attained. The Lean development proves the main theorem in the paper's normalization together with Smale's bound K(d)≤4K(d) \le 4, that K(d)K(d) does not decrease with dd, and that K(d)<4K(d) < 4 in every degree.

Novelty check

arXiv, 2026-11-16, abs:"mean value" AND abs:Smale AND abs:constant returned 3 records: this paper (2611.17373), Jatar and Ng's 2610.21472, which states the exact value as open, and a 2004 paper on mean value conjectures for rational maps. Within this registry 2026-09-03-smale-mean-value-k1 is a different and weaker result: it shows the constant exceeds 1, while this one determines its value. vibemathed records the result as solved on 6 November 2026, earlier than the arXiv posting used as the date here, probably from a Zenodo version; that date was not confirmed for this entry.

Caveats and known objections

Graded formal on the Palomar registration, which records trust level high for three theorems at a pinned commit. vibemathed's check read the Lean definitions of K(d) and its supremum line by line and found that they match the problem as Smale and Jatar and Ng pose it; Google DeepMind's Formal Conjectures states only the K = 1 form, so nothing written independently anchors this statement. The paper is unrefereed, and Hexagon labels it a mix of human-written and AI-generated text with human understanding of some parts. The proof gives no bound on the degree needed for K(d)K(d) to exceed a given value below 4. Autonomy is collaborative: the author writes that the work was done overwhelmingly by AI models but steered by him in a continuous critical dialogue, which could support ai-led, and the weaker grade wins when the reading is arguable. Model versions are not stated.

Machine-checked elsewhere

PalomarSmale's mean value constant equals 4

PALOMAR-2026-11-12-000015 · david44hello/paper_smale-mean-value-constant-equals-4 at 72511404d606 · checked 2026-11-16

ProvedSmale.P030.Rows.thm_1_1Smale.P030.kstar_eq_fourSmale.P030.tendsto_Kd_four

A Lean proof that typechecks against the recorded statement at a pinned commit under a declared axiom set, checked mechanically rather than by human review. It does not certify that a result is new or of research interest: the only filter on that is a language-model screen, and Palomar states that it adds no human editorial step.

Trust level high. The record states the result as answering Smale's question in the negative rather than proving his conjecture.

Also recorded at

vibemathedsmale-s-mean-value-constant

Nothing mechanically. It is a parallel listing of the same result, carrying its own verification label rather than an independent check.

Labels it Lean-checked with the statement unaudited, and records reading the headline definitions line by line without rebuilding the proof.

No independent check is recorded on this entry yet.

Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.

Or on GitHub: submit a check challenge the grade send a correction or send a pull request

Entry history (1 event)
  1. AddedEntered the registry graded Formally verified and Collaborative.

Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.

Graded formally verified for verification and collaborative for autonomy. What these mean.

Cite this entry

Plain text
whataifound.org. (2026). Smale's mean value constant shown to equal 4, with a Lean proof. whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2026-11-10-smale-mean-value-constant-four
BibTeX
@misc{whataifound-independent-2026-four,
  title        = {Smale's mean value constant shown to equal 4, with a Lean proof},
  author       = {{whataifound.org}},
  year         = {2026},
  howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
  note         = {Result by Independent. Verification: Formally verified. Autonomy: Collaborative.},
  url          = {https://whataifound.org/finding/2026-11-10-smale-mean-value-constant-four}
}

Related findings

← All mathematics findings in the registry