VibeMathedMath problems solved with AI

The van der Put-Singer conjecture: the inverse problem in difference Galois theory over C(x)C(x)

Let CC be an algebraically closed field of characteristic zero and K=C(x)K=C(x) with the shift σ(x)=x+1\sigma(x)=x+1. Every linear difference system σY=AY\sigma Y=AY with A∈GLn(K)A\in\mathrm{GL}_n(K) has a Picard-Vessiot (difference Galois) group, a linear algebraic group over CC. Van der Put and Singer proved that such a group GG always has a cyclic component group G/G∘G/G^\circ, and conjectured the converse: does every linear algebraic group GG for which G/G∘G/G^\circ is cyclic occur as a difference Galois group over C(x)C(x)?

Result
Proved(see note)
Status
Candidate (review pending)
AI contribution
AI-discovered
Method
Argument
Field
Difference Galois theory; inverse problem
Posed by
Marius van der Put and Michael F. Singer, Galois Theory of Difference Equations, Lecture Notes in Mathematics 1666 (1997), Chapter 3
Year posed
1997
Years open
29y
Solved
2026-11-08
Model
ChatGPT 6 with the MechMath Agent Team (MMAT)
Vendor
OpenAI; MMAT by the Key Laboratory of Mathematics Mechanization, AMSS, CAS
Collaborators
Ruyong Feng
Verification
Unreviewed
Publication
Preprint
Significance
25 / 100
Disclosed cost
—
Wikipedia
No dedicated article

What was actually shown

Theorem 1.1 of arXiv:2611.11397 (Ruyong Feng, v1 8 November 2026): for every closed subgroup G⊆GLn(C)G\subseteq\mathrm{GL}_n(C) with G/G∘G/G^\circ cyclic there are A∈G(K)A\in G(K), a Picard-Vessiot ring RR for σY=AY\sigma Y=AY and a fundamental matrix Z∈G(R)Z\in G(R) such that τ↦Z−1τ(Z)\tau\mapsto Z^{-1}\tau(Z) maps Autσ(R/K)\mathrm{Aut}_\sigma(R/K) onto G(C)G(C). Moreover A(x)=h+O(x−2)A(x)=h+O(x^{-2}) at infinity with h∈G(C)h\in G(C) semisimple. With the known necessity this gives: GG is a difference Galois group over C(x)C(x) if and only if G/G∘G/G^\circ is cyclic. The new ingredients are the construction of such an AA with a condition on its finite singularities, and a rigidity theorem for rational difference-invariant lines; the rest uses Propositions 1.21 and 3.3 of the monograph. The qq-difference analogue is not treated. A footnote records that the second assertion of the monograph's Proposition 3.2(2) is false in general, with counterexamples.

What the AI did

From the paper's abstract: "The original results in this paper were entirely generated by ChatGPT 6 and the MechMath Agent Team (MMAT) developed by the Key Laboratory of Mathematics Mechanization, AMSS, CAS. The author reviewed and simplified the initial proofs and verified the correctness of the results." The contributions section adds that the proofs were generated after the author supplied the reference materials; that they contained both new ingredients, the coefficient matrix with A(x)=h+O(x−2)A(x)=h+O(x^{-2}) and the rigidity theorem, and concluded by descent over a Picard-Vessiot ring; and that the author replaced the descent by a direct use of Proposition 1.21 of the monograph and simplified the proof of Lemma 3.2. The original AI-generated proofs are posted on the author's page.

Verification

Read by this site on 19 November 2026 in the TeX source of arXiv:2611.11397v1 (single author, Ruyong Feng, Academy of Mathematics and Systems Science, Chinese Academy of Sciences): the abstract with its AI statement, the introduction, Theorem 1.1, the contributions section, and the rigidity theorem with its supporting lemmas; the construction sections were skimmed. The proof was not refereed here and there is no formal proof. The monograph of van der Put and Singer was not opened: the conjecture (Chapter 3, page 35) is taken from the paper's citation, and it agrees with Conjecture 6.1 in the source of Feng's earlier paper arXiv:1708.07944, which states it as an if and only if and notes that only sufficiency was open. The file of original AI-generated proofs linked from the paper opens (checked, not read). The acknowledgements thank Michael Singer, who co-posed the conjecture, for suggestions; that is not a public confirmation. A web search found no earlier proof and no public expert reaction.

Sources

Submitted by MellowGander562 on

Changelog2 changes

Discussion