VibeMathedMath problems solved with AI

Smale's sixth problem for six bodies in the plane: finiteness of central configurations for generic masses

A configuration q=(q1,…,qn)q=(q_1,\ldots,q_n) of nn distinct points in the plane, carrying positive masses m1,…,mnm_1,\ldots,m_n, is a central configuration if there are λ∈R\lambda\in\mathbb{R} and c∈R2c\in\mathbb{R}^2 with λmi(qi−c)=∑j≠imimj(qi−qj)∣qi−qj∣3,i=1,…,n.\lambda m_i(q_i-c)=\sum_{j\neq i}\frac{m_i m_j(q_i-q_j)}{|q_i-q_j|^3},\qquad i=1,\ldots,n. Smale's sixth problem asks whether, for every choice of positive masses, the number of central configurations up to similarity is finite. Hampton and Moeckel (2006) proved it for n=4n=4, and Albouy and Kaloshin (2012) for n=5n=5 outside a codimension-two set of masses. Is it true for n=6n=6?

Result
Proved(see note)
Status
Partial result
AI contribution
AI-assisted
Method
Computation
Field
Celestial mechanics; central configurations
Posed by
Smale, Mathematical problems for the next century (Math. Intelligencer 20, 1998), Problem 6; earlier Chazy (1918) and Wintner (1941)
Year posed
1998
Years open
28y
Solved
2026-10-27
Model
Generative AI tools (unspecified)
Vendor
—
Collaborators
Tejasvi Singh Tomar
Verification
Unreviewed
Publication
Preprint
Significance
42 / 100
Disclosed cost
—
Wikipedia
No dedicated article

What was actually shown

Theorem 1.1: there is a nonzero polynomial F∈Z[m1,…,m6]F\in\mathbb{Z}[m_1,\ldots,m_6] such that for every positive mass vector with F(m)≠0F(m)\neq 0 the planar six-body problem has finitely many central configurations up to similarity. FF is not computed and its zero set is described only in part: the theorem says nothing when three masses are equal (so not for six equal masses) or when one mass is the sum of two others. Finiteness for all positive masses remains open, as the paper says, so this is a partial result on Smale's sixth problem: the six-body analogue of the generic five-body theorem of Albouy and Kaloshin. The proof follows their method: a curve of complex solutions is excluded through the valuations of the mutual distances at its points at infinity, with linear programs, Gröbner bases and interval arithmetic.

What the AI did

The paper's acknowledgement, in full: "Generative AI tools were used to write code, run analyses and draft text; the author directed the work and takes full responsibility for its content." No model or vendor is named. The proof is computer-assisted and formalized in Lean, so the code and analyses are a large part of it; the statement does not say which mathematical ideas, if any, came from a model. Hexagon labels the text a mix of human-written and AI-generated, with human understanding of all parts.

Verification

No independent mathematician has checked this yet. Unreviewed on this site's ladder: a Lean formalization exists, but that its headline avoids native_decide was not confirmed here. Checked here on 19 November 2026: the Hexagon record and 48-page PDF (hexagon:2611.00377v1) and the Lean framework archive of the Zenodo supplement (5,586 Lean files). The formal statement CC6Full.MainTheorem (CC6Full/Statement.lean) was read line by line: it is GenericFiniteness 6, a nonzero F∈Z[m1,…,m6]F\in\mathbb{Z}[m_1,\ldots,m_6] such that every positive mass vector with F(m)≠0F(m)\neq 0 has a finite set of central configurations representing all of them up to similarity, with the central configuration equation written out over C\mathbb{C}. That matches Theorem 1.1 and the generic form of Smale's problem. No sorry and no declared axiom were found. native_decide does occur, in 1,821 files, 462 of them in the import closure of the headline CC6Compose.V4.main_theorem_v4; the README explains these as the earlier version 2 and 3 routes, replaced in version 4 by kernel-checked twins, and reports that the headline depends only on propext, Classical.choice and Quot.sound. That report was not reproduced: nothing was rebuilt, and six generated certificate modules sit in archives of about 2 GB that were not downloaded. Nothing independent anchors the statement. Prior work was checked: Jensen and Leykin completed their computation only for n≤5n\le 5.

Sources

Submitted by MellowGander562 on

Changelog2 changes

Discussion