Documentary state: EDITORIALLY_ADMITTED. The manuscript has a protected editorial admission, not a theorem certification.
OAI-26 family 004: Hilbert's tenth problem over the rational numbers
The question in plain language
Given a polynomial with integer coefficients, can a computer always decide whether it has a solution in rational numbers? A rational number is a ratio of two integers. Hilbert's tenth problem over the rational numbers asks for one algorithm that works for every polynomial, even when the number of variables varies. The pinned October 6, 2026 manuscript claims there is no such algorithm. This is an announcement by its authors, not a proof independently certified by GCL.
The analogous decision problem for integer solutions is known to be undecidable. That fact alone does not settle the rational problem: permitting denominators changes the set of solutions, and a polynomial might have rational solutions without integer solutions.
A small example
The polynomial \(x^2-2\) has no rational zero, even though it has real zeros. By contrast, \(2x-1\) has the rational zero \(x=1/2\) but no integer zero. Both are readily decided examples. An impossibility result about one algorithm handling all polynomials is much stronger than proving that some individual equations are difficult.
The manuscript describes a way of building specially constrained polynomial families for which rational solvability would encode integer solvability. A correct encoding would be the key bridge; these simple examples do not establish it.
What is known
Main pinned paper: Hilbert's tenth problem over the rational numbers, October 6, 2026, PDF Theorem 1.1 (physical p. 3), claims that there is no algorithm which, given an arbitrary polynomial in an arbitrary finite number of variables with integer coefficients, decides whether it has a zero in \(\mathbb Q\). Its proof overview (§1.2, p. 5) describes reducing integer-solvability questions to a computable sequence of finite rational-solvability tests. The proposed machinery involves elliptic-curve indices, local comparisons, height bounds at selected primes, and a prime-pattern construction. Corollary 7.1 (p. 56) additionally claims a Turing-degree statement (\(0'\)) and restricted quartic and sums-of-squares-of-quadratics forms. These remain claims of the pinned manuscript.
Companion pinned paper: A pointwise 2-converse for elliptic curves with rational two-torsion, October 7, 2026, Theorem 1.1, claims that if an elliptic curve over \(\mathbb Q\) has a nontrivial rational point of order two and its full \(2^{\infty}\) Selmer group has corank zero or one, then the analytic and Mordell–Weil ranks equal that corank and the full Tate–Shafarevich group is finite. It does not claim the exact BSD leading-coefficient formula. The main Hilbert-ten paper invokes the companion as reference [48] in the proof chain around Lemma 3.4.
The key logical distinction is between a conditional local construction of rational points and a uniform reduction proving undecidability for every input. The source claims both; inspecting one local lemma is not a validation of the final reduction.
GCL verification state
The source is locked to openai/math@1581d4cbdd69a0faf34b2089614fe2a814316a6a. The 62-page Hilbert-ten PDF has Git blob b75dddd5e05503cc04a156e94bb15d6fadf8158e; its 91-page 2-converse companion has Git blob 1d35e83ccc541890903bfa978c1f0f73972843ca. Prior bounded source reads are recorded in the Family 004 review thread.
The source-specific audit of main-paper §§3.3–3.4 (PDF pp. 21–24) checked a finite-field reciprocity identity and, conditional on the stated matrix-rank and local Kummer hypotheses, the ensuing elimination step. The paper constructs an elliptic curve \(E_l\) and reports \(\dim_{\mathbb F_2}\mathrm{Sel}_2(E_l/\mathbb Q)=3\) with full rational two-torsion. The Kummer exact sequence then gives
The manuscript uses the full two-power Selmer-corank bound to invoke its pointwise companion, obtaining finiteness of \(\Sha\) under that claimed theorem. A perfect alternating Cassels–Tate pairing for finite \(\Sha\) then makes the dimension of \(\Sha[2]\) even; given the above equation it must be zero. Under these assumptions, the chosen 2-Selmer class \(\gamma\) is represented by a rational point. This checks a conditional chain of implications, not the upstream Selmer calculation or the companion's proof.
Specifically still unverified are the construction of a squarefree twist satisfying the matrix hypotheses, all exceptional-place local Kummer comparisons, the exact global Selmer dimension, and the later reduction from polynomial equations to rational-solvability tests. No theorem-level Lean correspondence, Comparator acceptance, MATHCERT certificate, or independent proof validation has been established.
Documentary state: DRAFT_UNACCEPTED. Semantic state: PINNED_MAIN_AND_COMPANION_STATEMENTS_CHECKED; CONDITIONAL_LEMMA_3_4_CHAIN_REVIEWED; GLOBAL_PROOF_UNVERIFIED. NO_CERTIFICATION.
What remains
Complete source-critical review of the final undecidability reduction, including the construction of the required prime patterns and the passage from finite rational tests to an integer zero. For Lemmas 3.3–3.4, verify the exact source hypotheses (especially equations (3.11), (3.15)–(3.17)), exceptional local Kummer calculations and all conditions for applying the pointwise 2-converse. Check the restricted-normal-form corollary separately.
For editorial acceptance, the issue is whether the source-qualified claims and limits here are faithful to the pinned documents, not whether GCL has proved Hilbert's tenth problem over \(\mathbb Q\). Require a critic tied to this exact draft, strict Markdown/MkDocs rendering, and an exact-hash protected editorial admission. Mathematical certification would require a substantially different independent proof review.
Sources and claim boundaries
- Pinned Hilbert-ten main PDF.
- Pinned pointwise 2-converse PDF.
- GCL Family 004 source-audit and critic evidence.
- OAI26-ED-01 work package.
Boundary: an auditable account of a proposed proof is not an independently verified proof. No published claim, source extraction or conditional algebraic check is promoted here to mathematical truth or certification.
Indexed primary manuscript paths:
- Hilbert’s tenth problem over the rational numbers —
preprints/Hilberts-tenth-problem-over-the-rational-numbers-October-6-2026/main.pdf - A pointwise 2-converse for elliptic curves with rational two-torsion —
preprints/A-pointwise-2-converse-for-elliptic-curves-with-rational-two-torsion-October-7-2026/paper.pdf