VibeMathedMath problems solved with AI

Rohrbach's conjecture on thin hh-bases of finite groups

Let GG be a finite group and h≥2h\ge2. A subset A⊆GA\subseteq G is an hh-basis if Ah=GA^h=G, that is, every element of GG is a product of hh elements of AA. Counting gives ∣A∣≥∣G∣1/h|A|\ge|G|^{1/h} for every hh-basis. Rohrbach asked in 1937 whether this is sharp up to a constant: for every fixed hh, is there a constant chc_h, independent of GG, such that every finite group GG has an hh-basis AA with ∣A∣≤ch∣G∣1/h|A|\le c_h|G|^{1/h}?

Result
Proved(see note)
Status
Candidate (review pending)
AI contribution
AI-assisted
Method
Argument
Field
Finite group theory; thin bases and product decompositions
Posed by
Hans Rohrbach, Anwendung eines Satzes der additiven Zahlentheorie auf eine gruppentheoretische Frage, Math. Z. 42 (1937), 538-542
Year posed
1937
Years open
89y
Solved
2026-11-12
Model
ChatGPT (version not stated)
Vendor
OpenAI
Collaborators
Yong Yang
Verification
Unreviewed
Publication
Preprint
Significance
28 / 100
Disclosed cost
—
Wikipedia
No dedicated article

What was actually shown

Theorem 1.2 of arXiv:2611.16938 (Yong Yang, v1 12 November 2026): for every hh there is chc_h such that every finite group GG has A⊆GA\subseteq G with Ah=GA^h=G and ∣A∣≤ch∣G∣1/h|A|\le c_h|G|^{1/h}; one may take ch=1+C(h−1)c_h=1+C(h-1). It follows from the stronger Theorem 1.1: there is an absolute constant CC such that for all α1,…,αh≥0\alpha_1,\dots,\alpha_h\ge0 with sum 11 there are A1,…,Ah⊆GA_1,\dots,A_h\subseteq G with G=A1⋯AhG=A_1\cdots A_h, ∣A1∣≤∣G∣α1|A_1|\le|G|^{\alpha_1} and ∣Ai∣≤C∣G∣αi|A_i|\le C|G|^{\alpha_i} for i≥2i\ge2. The paper presents this as completing the Modified Rohrbach Problem as well. The proof reduces to finite simple groups by the Kozma-Lev extension lemma and treats the groups of Lie type family by family, so it depends on the classification of finite simple groups. The constant is not explicit: 3(3/2)1203(3/2)^{120} for the exceptional families, enlarged by an unspecified amount for finitely many small groups and the sporadic groups.

What the AI did

From the paper's acknowledgements: "OpenAI's ChatGPT was used for mathematical exploration and editorial assistance. The author checked the final arguments and is responsible for the contents of the paper." The paper does not say which parts of the argument came from the model, or which version was used.

Verification

Read by this site on 19 November 2026 in the TeX source of arXiv:2611.16938v1 (single author, Yong Yang, Texas State University): the problem as stated in the introduction, Theorems 1.1 to 1.3, the whole argument (about 13 pages) and the AI sentence in the acknowledgements. The steps were followed and no error was noticed, but this is a reading, not a referee report: the cited inputs (the Burness-Li factorization used for the plus-type orthogonal groups, the Bruhat decomposition facts, Kozma and Lev for the alternating groups) were not opened, and there is no formal proof. Rohrbach's 1937 papers were not opened either; the statement is taken from the paper and agrees with the MathWorld entry on the conjecture, which already cites this preprint. A web search on 19 November found no public confirmation by another mathematician and no earlier proof for all finite groups.

Sources

Submitted by MellowGander562 on

Changelog2 changes

Discussion