{"id":"800d174f-4e1a-40f9-a4bd-49542ad21049","arxiv_id":"2607.03944","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For a smooth hypersurface of dimension n≥6, the Ulrich complexity is at least about n−1 (odd n), n−2 (even n), and at least n when n is even and the hypersurface is very general.","lead":"This paper proves a new lower bound on the Ulrich complexity of smooth hypersurfaces: the minimal rank of an Ulrich bundle grows at least linearly with the dimension. The bound improves the previous square-root estimate and is nearly linear for cubic hypersurfaces.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Very-general even bound depends on unpublished [LR2] Hodge lemma; odd and non-very-general bounds appear self-contained.","rationale":"The reader's weakest_assumption correctly identifies the load-bearing external dependency. I re-checked the proof of Lemma 4.1 and the subsequent combinatorial Claim 4.2: the resolution of i_*E is standard, the GRR computation (4.1) is correct, the reduction to Q-twisted Newton sums (4.4) is valid, and the partition-sum identity (4.8) with the estimates (4.14)–(4.19) are sound. The only non-self-contained input is the assertion that for very general even-dimensional X all algebraic classes in H^n(X;Q) are multiples of H^{n/2}, imported from [LR2, Lemma 4.1], a preprint by two of the three authors. Without that, the very-general even part of Theorem 1 does not follow; the odd and non-very-general even parts remain valid. A minor boundary issue exists for n=6, where the even-case formula calls for F(2) while F is defined only for m≥3; this should be fixed (e.g., define F(2)=4 or restrict the range to n≥8), but it does not affect the asymptotic linear bounds. Since the main proof is otherwise convincing, I recommend keeping the conditional verdict.","tokens_in":11106,"tokens_out":22090,"duration_ms":205194,"concrete_test":"Retrieve arXiv:2503.13396 and read [LR2, Lemma 4.1]. Verify that it states exactly that for a very general smooth hypersurface X of even dimension n≥6 and degree d≥3, every algebraic class in H^n(X;Q) lies in Q·H^{n/2}, and that its proof is self-contained or cites only standard results. If the lemma holds as stated, the dependency is benign; if it is absent or has additional hypotheses, the theorem should be restated with the very-general even bounds made conditional on [LR2, Lem. 4.1].","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 4.1 (and hence the very-general even-dimensional part of Theorem 1: uc(X)≥n for n≥44 and uc(X)≥F(n/2) for 6≤n≤42) requires, for even n, that every algebraic class in H^n(X;Q) be a multiple of H^{n/2}. The proof refers to [LR2, Lemma 4.1], an unpublished preprint by two of the three authors, and gives no statement or proof of this Hodge-theoretic input. If that lemma is false, incomplete, or carries extra hypotheses (e.g., a lower bound on n or a restriction on d), the very-general even bounds do not follow. The odd-dimensional and non-very-general even-dimensional lower bounds do not rely on this lemma; they follow from the Ulrich resolution and the hyperplane-section argument, and their combinatorial core (Claim 4.2, equations (4.6)–(4.19)) is sound.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a lower bound on Ulrich complexity of smooth hypersurfaces X⊂P^{n+1} of dimension n≥6 and degree d≥3. For odd n the bound is n−1 for n≥43 and min{2m,B(m)} with m=(n−1)/2 for small n; for even n it is n−2 (respectively n for very general X) with analogous small-n values. The proof works with the Chern character of an Ulrich bundle, twists by −uH to turn the character into a series with coefficients b_j, and then uses a partition expansion to show that if the rank r is below the threshold F(m), the Chern class c_{2m} of the twisted bundle is nonzero, contradicting r<2m. The even non-very-general bound is obtained by restricting to a hyperplane section.","tokens_in":11330,"tokens_out":7998,"duration_ms":80393,"significance":"If the result holds, it is the first linear lower bound on Ulrich complexity of hypersurfaces, substantially improving the previous sqrt(n+2)−1 bound of [BES]. The core of Section 4 is an explicit, parameter-free contradiction argument: the coefficient estimates in Lemma 2.3, the partition expansion (4.8), and the domination estimate (4.19) are all checkable and appear sound. The odd-dimensional and non-very-general even bounds are essentially self-contained. The stronger very-general even bound, however, depends on a Hodge-theoretic input imported from an unpublished preprint of two of the authors, so the paper as it stands is only conditional for that portion of the main theorem.","major_comments":[{"comment":"The very-general even-dimensional bounds uc(X)≥n (n≥44) and uc(X)≥F(n/2) (n≤42) require the assertion that for a very general even-dimensional hypersurface every algebraic class in H^n(X;Q) is a multiple of H^{n/2}. This is imported from [LR2, Lemma 4.1], an unpublished preprint by two of the three authors, and the lemma is neither stated nor proved in the present paper. If that lemma fails, or carries extra hypotheses on n or d, the very-general even part of Theorem 1 does not follow. Since this is load-bearing for part of the main claim, please include a complete statement and proof of the needed Hodge-theoretic fact, or reformulate Theorem 1 so that the very-general even bounds are explicitly conditional on it.","section":"§4, Lemma 4.1; Theorem 1 (even very-general case)"}],"minor_comments":[{"comment":"The last displayed formula ends with \"F(n/2). .\" — a stray double period, and the entire theorem lacks a closing period.","section":"Theorem 1 statement"},{"comment":"The sentence \"Set k=2sin (3.1)\" should read \"Set k=2s in (3.1)\".","section":"§4, proof of Theorem 1"},{"comment":"In the displayed definition of G_N(z), the product notation \"NY i≥1\" should be \\prod_{i=1}^N.","section":"Lemma 2.2 proof"},{"comment":"The statement that i_*: H^{2k}(P^{n+1};Q)→H^{2k}(X;Q) is an isomorphism for all k≠n/2 is not immediate for k>n/2. It follows from hard Lefschetz, but a one-sentence justification (or a precise reference) would be useful.","section":"§4, Lemma 4.1"},{"comment":"The term \"very general\" is used but never defined. Please specify the precise countable union of proper loci in the moduli of degree-d hypersurfaces that is excluded.","section":"Introduction / Theorem 1"},{"comment":"Ref. [LR2] is cited for a fact that is load-bearing for part of the theorem. If the preprint is not yet published, its statement should be reproduced in the paper or an appendix should give the proof; at minimum, update the reference if it has appeared.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The main concern is the unpublished-dependency in the very-general even case. The core Section 4 argument is sound and self-contained for the odd and non-very-general even bounds, so the paper is promising. However, as submitted, the strongest theorem is partly conditional on a lemma from an unpublished preprint by two of the authors; for a journal publication this should either be proved in the paper or the statement should be made explicitly conditional."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing you should know: this paper delivers what it says. It proves the first lower bound on Ulrich complexity of smooth hypersurfaces that is linear in the dimension, improving the known sqrt(n+2)-1 from [BES] to roughly n. The main strategy is clean: use the Ulrich resolution to compute the Chern character of the twisted bundle, then show that a rank r < F(m) would force certain intersection numbers to have alternating signs while also vanishing for rank reasons. The analytic ingredient, Lemma 2.3, bounding the coefficients of the even function d/(e^{uz}+...+e^{-uz}), is a nice piece of elementary analysis, and the partition-sum domination in Claim 4.2 is explicit. I checked the key steps: the recurrence (4.6), the expansion (4.8), and the geometric-series bound (4.19) all hold. For odd-dimensional hypersurfaces and for non-very-general even-dimensional ones, the proof is self-contained and convincing.\n\nThe soft spot is exactly where the stress-test note puts it. The very-general even case of Theorem 1 (uc(X) >= n for n>=44, and F(n/2) for n<=42) depends on Lemma 4.1's assertion that every algebraic class in H^n(X;Q) on a very general even-dimensional hypersurface is a multiple of H^{n/2}. That fact is imported from [LR2, Lemma 4.1], an unpublished preprint by two of the three authors, and the paper neither states the lemma nor sketches its proof. If that lemma is true, the argument goes through; if it has extra hypotheses or a gap, the very-general even bounds do not follow. The odd-dimensional and non-very-general even bounds do not rely on it. This is worth fixing before publication: either prove the Hodge statement or at least state it precisely with the hypotheses.\n\nThe novelty is real: this is not a reparametrization of the [BES] bound. The relation to the authors' related preprint [LR2] is mentioned in passing, but the novelty boundary between this paper and [LR2] should be made explicit. That is a presentation issue, not a mathematical one.\n\nBottom line: this paper deserves a serious referee. I would send it to someone who can check the Hodge-theoretic input and the combinatorial estimates. If the [LR2] lemma is supplied, it should be published as is.","headline":"Genuinely new linear lower bound for Ulrich complexity; the core argument is convincing, but the very-general even-dimensional case rests on an unpublished Hodge lemma from the authors' own preprint.","tokens_in":11804,"tokens_out":2078,"would_cite":true,"duration_ms":20260,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14J70","14J60","14F06"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes that the Ulrich complexity of any smooth hypersurface of dimension n grows at least linearly in n, with explicit rank bounds that for large dimensions read n−1 (odd n) and n−2 (even n).","keywords":["Ulrich bundles","Ulrich complexity","hypersurfaces","Chern character","lower bound","algebraic Hodge classes","Newton identities","partition combinatorics"],"falsifier":"Exhibit a smooth hypersurface of dimension n≥6 carrying an Ulrich bundle of rank strictly below the claimed bound (e.g., rank ≤ n−3 for n≥44, or rank < F(m) in the small range), or find a very general even-dimensional hypersurface with an algebraic class in H^n not proportional to H^{n/2}. Either would refute the corresponding part of Theorem 1.","tokens_in":11015,"feed_emoji":"📏","tokens_out":13077,"duration_ms":114147,"temperature":0.7,"pith_summary":"The paper proves the first linear lower bound on the Ulrich complexity of smooth hypersurfaces: the minimal possible rank of an Ulrich bundle on an n-dimensional smooth hypersurface grows at least as a constant multiple of n, and in large dimensions is n−1 (n odd) or n−2 (n even). The previously known bound was only on the order of sqrt(n). This matters because Ulrich bundles are a powerful tool in the study of projective varieties, and their minimal ranks—the Ulrich complexity—are expected to be large; linear lower bounds are a concrete step toward exponential conjectures. The proof combines a character computation via Grothendieck-Riemann-Roch with sharp estimates on a trigonometric power series and a combinatorial domination argument over integer partitions.","feed_headline":"Ulrich complexity of hypersurfaces grows linearly with dimension","feed_subtitle":"New theorem gives explicit rank lower bounds n−1 and n−2, improving the old sqrt(n+2)−1 result.","key_machinery":"The proof computes the Chern character of an Ulrich bundle on a hypersurface. Grothendieck-Riemann-Roch and the linear resolution of an Ulrich sheaf force the character to equal rd/(1+e^{-H}+...+e^{-(d-1)H}) (under a Hodge-structure hypothesis for even very general X). Twisting by the rational multiple −(d−1)H/2 yields a generating function f(z) whose Maclaurin coefficients b_j satisfy sharp two-sided estimates. Newton's identities convert these estimates into a recursion for the intersection numbers J_s = c_{2s}(F)H^{n−2s}. The combinatorial core is Claim 4.2: expanding J_m as a sum over partitions of m, the 'full' term A_m dominates the sum of all other terms in absolute value whenever r <","core_discovery":"The paper's central claim is Theorem 1: if X is a smooth hypersurface of dimension n≥6 and degree d≥3, then any Ulrich bundle on X has rank at least n−1 when n is odd and n≥43, at least n−2 when n is even and n≥44, and at least n when n is even, n≥44, and X is a sufficiently general (very general) member of its moduli space. For smaller dimensions, the lower bound is the explicit number F(m) = min{2m, B(m)} with m = ⌊n/2⌋ (or (n−2)/2), where B(m) is a closed-form rational expression in d and m. These are the first linear lower bounds on the Ulrich complexity of hypersurfaces, improving the prior general inequality uc(X) ≥ sqrt(n+2) − 1. In particular, for smooth cubic hypersurfaces the bound","pith_inferences":["If the Hodge-class statement for very general even-dimensional hypersurfaces (that all algebraic classes in H^n are multiples of H^{n/2}) were proved without genericity assumptions, the stronger n-bound would hold for all even-dimensional hypersurfaces; the current restriction argument yields only n−2 for non-very-general X.","The same partition-domination technique could be exported to other settings where the Chern character of a special bundle is governed by a product formula, e.g., Ulrich bundles on complete intersections; the required input is only a two-sided estimate like the one in Lemma 2.3.","The numerical cutoffs (n≥43/44, m≥21 for F(m)=2m) are artifacts of the chosen estimates; tightening the coefficient bounds in Lemma 2.3 could lower these cutoffs and yield exact 2m thresholds for smaller dimensions.","The theorem does not settle the exponential-rank conjectures for hypersurfaces, but shows that any such lower bound must kick in above a linear growth rate; a natural next test would be to search for Ulrich bundles of rank between n and 2^{n/2}."],"forward_implications":["For every smooth hypersurface of dimension n≥44, the Ulrich complexity is at least n−2, regardless of degree, so the minimal rank of an Ulrich bundle grows linearly with dimension.","For very general even-dimensional hypersurfaces, the lower bound is n for n≥44, two higher than the universal even bound n−2.","For smooth cubic hypersurfaces, the bounds are n−1 (odd n), n−2 (even n), and n (even very general); since Ulrich ranks on cubics are divisible by 3, the effective lower bound is the first multiple of 3 above these values.","For smaller dimensions the bound is the explicit function F(m) = min{2m, B(m)}, which for many low degrees d equals 2m (as tabulated), giving exact rank thresholds that depend only on degree and dimension.","The theorem provides the first linear lower bound, improving the previous general bound uc(X) ≥ sqrt(n+2) − 1."],"fun_headline_variants":["Ulrich complexity of hypersurfaces scales linearly with dimension","New rank lower bounds for Ulrich bundles on hypersurfaces","First linear bounds on hypersurface Ulrich complexity","Explicit lower bounds for Ulrich bundles on smooth hypersurfaces","Ulrich complexity of hypersurfaces: linear growth in dimension"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The proof of the stronger 'very general even' case assumes that for a very general even-dimensional smooth hypersurface, every algebraic cohomology class in middle degree is a rational multiple of H^{n/2}; this is imported from an unpublished companion preprint, and if it fails that part of the theorem collapses, though the odd-dimensional and non-very-general even bounds remain valid.","fun_headline_variants_meta":{"raw":{"variants":["Ulrich complexity of hypersurfaces scales linearly with dimension","New rank lower bounds for Ulrich bundles on hypersurfaces","First linear bounds on hypersurface Ulrich complexity","Explicit lower bounds for Ulrich bundles on smooth hypersurfaces","Ulrich complexity of hypersurfaces: linear growth in dimension"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000716,"raw_usage":{"total_tokens":2970,"prompt_tokens":575,"completion_tokens":2395,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":319,"completion_tokens_details":{"reasoning_tokens":2315}},"tokens_in":319,"tokens_out":2395,"duration_ms":14838,"temperature":1.0,"reasoning_tokens":2315,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T08:45:46.634472+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a smooth hypersurface of dimension n≥6 carrying an Ulrich bundle of rank strictly below the claimed bound (e.g., rank ≤ n−3 for n≥44, or rank < F(m) in the small range), or find a very general even-dimensional hypersurface with an algebraic class in H^n not proportional to H^{n/2}. Either would refute the corresponding part of Theorem 1.","supporting_citations":[],"review_version":2}