{"id":"6a6cbfa1-cb2a-435a-8c12-1e1c98b90e5d","arxiv_id":"2502.05528","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper proves new splitting theorems for p-adic differential modules over relative polyannuli and shows that the Christol-Mebkhout and Dwork definitions of p-adic exponents coincide.","lead":"This mathematics paper proves new decomposition theorems for certain p-adic differential equations on tube-like spaces, extending earlier results that worked only for simple base spaces. It also settles a long-standing question by showing that two different definitions of the main invariant, used by different research groups, actually agree.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Corollary 4.12 is not justified for non-strict Berkovich affinoid neighborhoods, so the proof of Theorem 4.18 needs a strictness hypothesis or a new local argument.","rationale":"The reader's weakest assumption was that Corollary 4.12 might fail when K is neither discrete nor algebraically closed. My concern is sharper: even for algebraically closed K, the statement of Corollary 4.12 fails for non-strict Berkovich spaces, which the paper's notation and definitions explicitly allow. The issue is not the cited finiteness results themselves but the function of strictness: a finite etale map to the strict unit polydisc forces the domain algebra to be strict, so non-strict affinoid points cannot be handled this way. This is load-bearing because the entire proof of Theorem 4.18 passes through this local reduction; without it, the decomposition on X0 is not established for such points. I do not claim the theorem is false: the statement may survive with a strictness hypothesis or with another reduction that treats non-strict Shilov points directly, and Proposition 4.9 already covers one-point Shilov cases. Therefore a conditional acceptance requiring the author to state the intended convention and close this gap is appropriate. The concrete test is a single, decisive computation: the non-strict disc over C_p at a radius outside |K^*|. This test settles whether the concern lands, because if the disc has no finite etale neighborhood map to B^1, the proof of Corollary 4.12 cannot hold in the stated generality.","tokens_in":49761,"tokens_out":51119,"duration_ms":548839,"concrete_test":"Take K = C_p, choose rho > 0 with rho notin |K^*| (for instance rho = |p|^{1/2}), and let X = M(K<rho^{-1}T>) be the closed disc of radius rho. Check whether the Shilov/Gauss point of X has any affinoid neighborhood U admitting a finite etale morphism U -> B^1_K. Since a finite etale cover of the strict algebra K<s> is strict, and every neighborhood of the Shilov point in this non-strict disc is non-strict, the expected answer is no. If no such U exists, Corollary 4.12 fails for non-strict smooth rigid spaces and Theorem 4.18 must either restrict to strict rigid spaces or find a different local reduction.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The proof of Theorem 4.18 reduces the general smooth base to the one-point Shilov boundary case by applying Corollary 4.12 at every point of X0: a finite etale morphism U -> B^m from an affinoid neighborhood to the unit polydisc. Corollary 4.12 is derived from Proposition 4.10, whose proof relies on R^o being topologically of finite type over O_K, citing [BGR84] and [Ach15]. But the paper works throughout with Berkovich spectra and allows non-strict affinoid algebras K<rho^{-1}s> for arbitrary rho in R_{>0}. If rho is not in |K^*|, which can happen even for algebraically closed K (e.g. K = C_p, rho = |p|^{1/2}), the closed disc M(K<rho^{-1}T>) is a smooth non-strict affinoid space. At its Shilov point, every affinoid neighborhood inside the disc has outer radius rho and is non-strict. A finite etale map to the strict unit polydisc B^m would make O(U) a finite module over a strict algebra, hence strict. Thus no such U exists, and Corollary 4.12 is false in the Berkovich setting the paper appears to use. If 'rigid space' was intended in the Tate/strict sense, that convention must be stated explicitly, because Definition 1.2 and Lemma 1.7 deliberately allow non-strict radii. Without either a proof of Corollary 4.12 for non-strict spaces or a strictness hypothesis on X, the main theorem's reduction is not complete.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a relative version of the p-adic Fuchs theorem for (logarithmic) ∇-modules over relative polyannuli X×A^n_K(I) satisfying the Robba condition. It introduces p-adic exponents for modules with relative connections, proves their well-definedness, studies their behavior under finite étale pushforwards, and proves a Galois descent. The main result, Theorem 4.18, asserts a direct-sum decomposition by exponents with non-Liouville differences for log-∇-modules over a connected smooth rigid base, generalizing Shiho's generization. Section 5 gives two generalized Fuchs theorems for Σ-semi-constant and \"ξ-constant on the base\" modules. The appendix proves that the set of exponents is exactly one weak equivalence class and that Christol-Mebkhout and Dwork exponents coincide.","tokens_in":50068,"tokens_out":21881,"duration_ms":213846,"significance":"If the strictness gap in §4.2 is repaired, the paper is a valuable contribution: Theorem 4.18 is a genuine relative p-adic Fuchs theorem over bases without one-point Shilov boundary, the pushforward and Galois-descent techniques are clean and reusable, and the appendix resolves two folklore questions (weak equivalence of exponents, CM vs Dwork definitions). The proofs are mostly well-structured, and the paper is honest about limitations in Remark 4.11. However, the main theorem currently rests on a local étale-structure statement that is not valid in the stated non-strict Berkovich setting.","major_comments":[{"comment":"Proposition 4.10 is not proved for the non-strict K-affinoid algebras admitted by Definition 1.2. The proof invokes [BGR84, Thm. 6.4.3/1] and [BGR84, Cor. 6.4.3/6], which give topological finite generation of R° over O_K for strictly affinoid algebras; for R=K⟨ρ^{-1}s⟩ with ρ∉|K^*| (possible over algebraically closed K, e.g. K=C_p and ρ=2 for odd p), R° is generally not topologically of finite type over O_K, so the presentation argument fails. Concretely, the Shilov point of M(R) has no affinoid neighborhood admitting a finite étale morphism to the strict unit polydisc, because such a neighborhood would have strict algebra, whereas every affinoid neighborhood of that point is the whole non-strict disc. Thus Corollary 4.12 is false in the stated Berkovich setting.","section":"§4.2, Proposition 4.10"},{"comment":"Theorem 4.18 depends on Corollary 4.12 by applying it to every point of X_0=X-D in the algebraically closed case. Since Corollary 4.12 fails for non-strict smooth rigid spaces (as explained above), the asserted reduction to the one-point Shilov boundary case is incomplete for such X. The theorem should be restricted to strictly K-analytic spaces (with an explicit convention, since Definition 1.2 and Lemma 1.7 deliberately use arbitrary radii), or a non-strict analogue of the local finite étale cover statement must be supplied. Without this, the main advertised decomposition is not established.","section":"§4.4, Theorem 4.18"}],"minor_comments":[{"comment":"The proof of Lemma 5.1 says \"Since the set of eigenvalues of A is (NLD)\", but the hypothesis is (NID). The argument should use that λ_i-λ_j-n ≠ 0 for n∈Z\\{0} because differences are non-integer; (NLD) alone would not exclude integer differences.","section":"§5.1, Lemma 5.1"},{"comment":"Several results (Propositions 2.4, 2.6, 2.10, 2.11, Theorem 2.13) are dismissed as \"the same as [Wan24]\". Since these statements are new relative versions over B^m_γ × A^n_K(I), the paper should either spell out the modifications or state explicitly that the cited proofs carry over verbatim with the relative variables treated as additional annulus directions.","section":"§2.1"},{"comment":"The notation ^kA, ^kN, ^k sgn is hard to parse; for instance, S_{k,B}=S^{σ_k}_{k,A}t^{-ksgn kN} should be written with the index k clearly separated from the step-k objects, e.g. using A^{(k)}, N^{(k)}, ε^{(k)}.","section":"Appendix A.1, Proposition A.3"},{"comment":"There are minor typos: \"thr n-th direction\" in Proposition 4.9 proof, \"boudary\" in Proposition 2.23 proof, \"veriry\" in Lemma 3.2 proof, and \"definiton\" in the introduction.","section":"Various"},{"comment":"The density statement \"R[t^{-1},t] is dense in K[[s/ρ]]^{an}⟨ρ'/t,t/ρ'⟩\" is used without proof or reference; a short justification (or a citation to the definition of analytic elements) would improve readability.","section":"§5.0, Lemma 5.4"},{"comment":"In the proof of Corollary 4.12 for dagger spaces, the appeal to [Vez18, Prop. 2.15] should state the exact statement that finite étale morphisms between completions extend to fringes, as Remark 4.13 notes, so that the étale morphism U→B^m lifts to the dagger category.","section":"§4.2, Corollary 4.12 (dagger case)"}],"recommendation":"major_revision","confidential_remarks":"The main concern is the strictness gap in §4.2. If the author intends Berkovich spaces with non-strict affinoids, the main theorem is currently unproved; if strictness is imposed, the paper's scope is narrower but still new. I would encourage the author to state a clear convention and adjust the statements. The reliance on the author's own [Wan24] is acceptable but should be made more explicit in the proofs. The appendix is a solid contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper has two genuinely useful pieces. Appendix A.1 proves that a multiset weakly equivalent to an exponent is again an exponent (Conjecture A.1), and Appendix A.2 proves the CM/Dwork exponent definitions coincide up to sign. These are clean, self-contained, and checkable results; I'd trust them.\n\nThe main body proves a broader relative Fuchs theorem than Shiho's, dropping the one-point Shilov boundary assumption on the base. Theorem 4.18 is the headline: an absolute logarithmic connection on a relative polyannulus with non-Liouville exponent differences decomposes into exponent pieces. The proof strategy is sensible: push forward to the unit polydisc via finite etale maps, decompose there, then descend.\n\nThe stress-test note is right, though. Corollary 4.12, the local finite etale cover to the strict unit polydisc B^m, is not justified for non-strict Berkovich affinoids. The paper explicitly allows non-strict algebras in Definition 1.2 (e.g., K⟨ρ^{-1}s⟩ for ρ not in |K^*|), and a finite etale map U→B^m would force O(U) to be finite over a strict algebra, hence strict. At the Shilov point of a non-strict disc, every affinoid neighborhood is non-strict, so no such U exists. Thus the reduction in Theorem 4.18 currently works only if the base is strict Tate rigid, which is never stated, or if a strictness hypothesis is added. This is a load-bearing gap, not cosmetic.\n\nOther issues are minor. The paper frequently says 'the proof is the same as [Wan24]' for factorization lemmas; that is acceptable because Wan24 is published, but it makes the paper less self-contained. Lemma 5.1 states NID but its proof uses NLD—the statement's condition is the correct one, so it's a typo. The relationship to Kedlaya's relative theory could be spelled out more, but that's not damaging.\n\nOverall: the appendix alone justifies referee time, and the main theorem is likely correct in the intended strict setting. A serious referee should ask for a precise convention about strictness, plus the Lemma 5.1 typo fix. This is not a desk reject; it deserves peer review.","headline":"Useful appendix results and a genuinely broader relative Fuchs theorem, but the main theorem's local finite-etale-cover step needs a strictness hypothesis or a new argument.","tokens_in":711,"tokens_out":1708,"would_cite":true,"duration_ms":82496,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["12H25","14G22"],"pacs":[],"model":"deepseek-v4-flash","headline":"p-adic connections decompose when exponents differ non-Liouville","keywords":["p-adic Fuchs theorem","relative polyannuli","Robba condition","logarithmic connections","p-adic exponents","rigid analytic spaces","Galois descent","weak equivalence of exponents"],"falsifier":"On the relative polyannulus $B^{1}$ × $A^{1}$_{Q_p}(]0,1[) for odd p, take a rank-2 relative ∇-module whose connection matrices have diagonal exponents 0 and 1/2 but with non-constant off-diagonal entries depending on the base coordinate, in a way that is not conjugate to a diagonal matrix over the relative annulus ring. Compute the intrinsic radius of convergence at the Gauss point; if such a module satisfies the Robba condition while remaining non-split, the canonical decomposition guaranteed by Theorem 4.18 would fail, refuting the theorem.","tokens_in":49530,"feed_emoji":"🔢","tokens_out":15120,"duration_ms":129196,"temperature":0.7,"pith_summary":"This paper proves a p-adic analogue of the classical Fuchs theorem for logarithmic connections on relative polyannuli: if a locally free module with an integrable connection satisfies the Robba condition and its p-adic exponents have non-Liouville differences, then the module decomposes canonically as a direct sum of submodules with constant exponents. This generalizes the 'generization' proposition of [Shi10], which required the base space to have a single-point Shilov boundary, to arbitrary connected smooth rigid bases. The paper also resolves two long-standing questions about exponents: the set of p-adic exponents of a module forms exactly one weak equivalence class, and the two standard definitions of exponent (from [CM97] and [Dwo97]) coincide. The proof relies on reducing the base to a unit polydisc via finite étale covers and then applying Galois descent.","feed_headline":"p-adic connections decompose when exponents differ non-Liouville","feed_subtitle":"New theorem works for any smooth rigid base, not just those with a one-point boundary.","key_machinery":"The argument is carried by three mechanisms. (1) A relative version of the Robba condition and p-adic exponents for ∇-modules over X × A^n(I) relative to X, defined pointwise and shown to be well-defined on connected bases by reduction to curves. (2) Pushforward of (relative and absolute) ∇-modules along finite étale morphisms, which preserves the Robba condition, exponents, and the uniqueness of decompositions with respect to Liouville partitions (Propositions 3.4, 3.7, Corollary 3.11). (3) A local structure theorem (Corollary 4.12) asserting that every smooth rigid or dagger space over an algebraically closed field is locally a finite étale cover of the unit polydisc, combined with Galois descent (Proposition 4.17) to pass back to arbitrary K. For the special base cases in Section 5, the machinery also includes the p-adic Birkhoff factorization of [Chr07].","core_discovery":"The central discovery is Theorem 4.18: let K be a complete nonarchimedean field of mixed characteristic, X a connected smooth rigid space, and P a logarithmic ∇-module over the relative polyannulus X × A^n_K(I) satisfying the Robba condition with exponent A. If the entries of A have p-adic non-Liouville differences, then P admits a unique direct sum decomposition P = ⊕_{λ ∈ (Z_p/Z)^n} P_λ, where each P_λ has exponent identically equal to λ. This is the p-adic Fuchs theorem for relative polyannuli in the absolute logarithmic case. The appendix additionally proves that the set of exponents of a Robba-condition module on a polyannulus is exactly one weak equivalence class, and that the two standard exponent constructions coincide.","pith_inferences":["If the announced theorem for relative connections over a base with one-point Shilov boundary (mentioned in Remark 4.19) is established, the pushforward technique of this paper would likely extend the Fuchs theorem to bases without any chosen differential structure on the base.","The local reduction to the unit polydisc suggests a strategy for proving relative p-adic Fuchs theorems for other classes of connections (e.g., irregular or with more general singularities) by first solving the problem on the polydisc and then gluing.","The single weak-equivalence-class property may simplify algorithmic computation of p-adic exponents in rigid cohomology: any weakly equivalent candidate multiset can be used as a working exponent for symbolic manipulation."],"forward_implications":["A log-∇-module over a relative polyannulus with non-Liouville exponent differences has a canonical eigen-decomposition, so its monodromy is described by constant exponents on each summand.","The generization proposition of [Shi10], previously restricted to bases with one-point Shilov boundary, now holds for arbitrary connected smooth rigid bases.","The coincidence of the two exponent definitions means results proved via one construction transfer automatically to the other.","The weak-equivalence result closes a gap in the literature: a multiset weakly equivalent to an exponent is again an exponent, so the set of exponents is a single weak equivalence class for modules on (relative) polyannuli."],"supporting_citations":[{"why":"Provides the generization proposition for bases with one-point Shilov boundary that the paper generalizes.","marker":"[Shi10]"},{"why":"The author's prior work establishing the high-dimensional absolute Fuchs theorem and the definition of exponents used as the foundation for the relative setting.","marker":"[Wan24]"},{"why":"Supplies the standard framework for p-adic differential equations, including the Robba condition and Dwork's method.","marker":"[Ked10]"},{"why":"Defines one of the two exponent constructions and proves the original one-dimensional Fuchs theorem.","marker":"[CM97]"},{"why":"Gives the other exponent construction and a proof of the Fuchs theorem; the appendix proves its equivalence with [CM97].","marker":"[Dwo97]"},{"why":"Contributes the algebraic geometry result used in Proposition 4.10 to construct finite étale covers of the unit polydisc.","marker":"[Ach17]"}],"fun_headline_variants":["Generalized p-adic Fuchs theorem for relative polyannuli","Decomposition theorem for p-adic connections on relative polyannuli","Non-Liouville exponents yield direct-sum decomposition of p-adic connections","Exponents with non-Liouville differences split p-adic connections","Robba-condition modules decompose under non-Liouville exponents"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof that every smooth rigid space has an affinoid neighborhood admitting a finite étale map to the unit polydisc (Corollary 4.12) requires the ring of topologically bounded elements to be topologically of finite type over the valuation ring of the base field; the author notes this can fail when the field is neither discrete nor algebraically closed, so the main theorem inherits this dependence.","fun_headline_variants_meta":{"raw":{"variants":["Generalized p-adic Fuchs theorem for relative polyannuli","Decomposition theorem for p-adic connections on relative polyannuli","Non-Liouville exponents yield direct-sum decomposition of p-adic connections","Exponents with non-Liouville differences split p-adic connections","Robba-condition modules decompose under non-Liouville exponents"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00158,"raw_usage":{"total_tokens":6258,"prompt_tokens":857,"completion_tokens":5401,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":473,"completion_tokens_details":{"reasoning_tokens":5312}},"tokens_in":473,"tokens_out":5401,"duration_ms":34738,"temperature":1.0,"reasoning_tokens":5312,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T18:57:38.298821+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On the relative polyannulus $B^{1}$ × $A^{1}$_{Q_p}(]0,1[) for odd p, take a rank-2 relative ∇-module whose connection matrices have diagonal exponents 0 and 1/2 but with non-constant off-diagonal entries depending on the base coordinate, in a way that is not conjugate to a diagonal matrix over the relative annulus ring. Compute the intrinsic radius of convergence at the Gauss point; if such a module satisfies the Robba condition while remaining non-split, the canonical decomposition guaranteed by Theorem 4.18 would fail, refuting the theorem.","supporting_citations":[],"review_version":1}