{"id":"ee5b497d-e6ac-48f1-b820-928d3da9e0b1","arxiv_id":"2507.09345","paper_version":1,"verdict":"CONDITIONAL","confidence":"LOW","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The general double cover of P^3 branched along a surface of degree 4, 6, or 8 carries a stable rank 2 Ulrich bundle, with moduli components of dimension 5, 6, and 0.","lead":"This thesis proves that some double coverings of 3D projective space admit the smallest possible Ulrich bundles, rank 2, exactly when the branch surface has degree 4, 6, or 8. These are special vector bundles with maximally simple cohomology, and the results give new examples and moduli-space dimensions for a long-standing existence question in algebraic geometry.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Existence of rank-2 Ulrich bundles on general double solids reduces to a polynomial genericity statement for b=p_0^2+p_1p_2+p_3p_4; for m=4 the dimensional estimate in the Introduction is not by itself a proof, and the differential rank must be checked.","rationale":"The reader's weakest assumption identifies exactly the step I consider load-bearing: the equivalence in Proposition 3.56 reduces the existence of rank 2 Ulrich bundles on general double solids to the representation of the general branch polynomial as p_0^2+p_1p_2+p_3p_4. The Introduction says this follows from 'dimensional estimates on polynomials', but the full argument in Section 3.2.4 is not present in the excerpt. My concern is not that the statement is false; rather, the supplied evidence is incomplete at the precise point where the main existence theorem rests. The proposed check is concrete and would settle the matter: surjectivity of the differential is a standard sufficient condition for dominance of a polynomial map between affine spaces of the relevant dimensions, so computing the ranks of the displayed multiplication maps for random tuples either certifies the genericity claim or exposes a genuine gap. Since my analysis reinforces rather than overturns the reader's conditional assessment, the verdict should remain unchanged.","tokens_in":977,"tokens_out":912,"duration_ms":176333,"concrete_test":"Run a Macaulay2 computation over Q (or over a large finite field of characteristic not 2) for m=2,3,4. Let S=QQ[x_0..x_3], choose five random elements p_0,...,p_4 in S_m, and compute the rank of the linear map S_m^5 -> S_{2m} sending (h_0,...,h_4) to 2p_0h_0+p_2h_1+p_1h_2+p_4h_3+p_3h_4. If for m=4 the rank equals 165, and for m=2,3 the ranks equal 35 and 84 respectively, then the differential is surjective at a point, so Φ is dominant and the genericity statement in Proposition 3.56 holds. If any of these ranks falls short, the corresponding existence assertion in Theorem 3.45 requires an additional argument showing the image still contains an open set, or it must be weakened.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of Theorem 3.45 — that the general double cover of P^3 branched in degree 4, 6, or 8 admits a stable rank 2 Ulrich bundle — is reduced, via Proposition 3.56, to proving that for m=2,3,4 the general degree-2m polynomial b in four variables lies in the image of the map Φ(p_0,...,p_4)=p_0^2+p_1p_2+p_3p_4. The only justification visible in the provided text is the phrase 'dimensional estimates on polynomials' in the Introduction. That estimate is necessary but not sufficient for dominance. For m=4 the source has dimension 5·C(7,3)=175 and the target C(11,3)=165, so a crude dimension count does not rule out dominance, but it also does not establish it. The correct and decisive condition is whether the differential of Φ is surjective at a general point. The image of dΦ is the degree-8 part of the ideal (p_0,p_1,p_2,p_3,p_4) generated by five general quartics in k[x_0,x_1,x_2,x_3]. If this is not all of S_8 for a general tuple, then the general b has no representation of the required form, and the existence assertion for m=4 — the Calabi-Yau case — fails. This is not an internal inconsistency of the paper, but it is the load-bearing step that the supplied text does not allow an independent reader to certify.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This thesis studies Ulrich sheaves on finite covers of projective space, concentrating on double covers. It proves a structural characterization (Propositions 3.54 and 3.56): for a smooth double cover f:X→P^n with Pic(X)=ZH and char k≠2, a rank-2 Ulrich sheaf exists if and only if the branch equation b of degree 2m can be written as b=p_0^2+p_1p_2+p_3p_4 with homogeneous p_i of degree m, and if and only if there is a curve Y⊂X mapped isomorphically to a complete intersection of two degree-m hypersurfaces. The paper then asserts (Theorem 3.45) that the general double cover of P^3 branched along a surface of degree 2m=4,6,8 admits stable rank-2 Ulrich bundles, and that the corresponding moduli components have reduced dimension 5, 6, and 0 respectively. It also constructs higher-rank Ulrich bundles on these double solids via extensions and deformations, studies the action of the covering involution, and analyzes restrictions to hyperplane sections and low-degree hypersurfaces.","tokens_in":65084,"tokens_out":7280,"duration_ms":95328,"significance":"If correct, the paper would determine the minimal Ulrich rank for three families of threefolds — the quartic, sextic, and octic double solids — and the octic case would be a comparatively rare Calabi-Yau example. The if-and-only-if criterion relating rank-2 Ulrich bundles to the polynomial decomposition b=p_0^2+p_1p_2+p_3p_4 is a genuine structural equivalence and is a valuable contribution in itself. The manuscript is also careful in its treatment of the Hartshorne-Serre correspondence in families and deformation theory, and it explicitly records a suspected gap in a cited argument (Remark 4.30). The main caveat is that the key algebraic genericity statement underlying Theorem 3.45 is announced in the introduction with a dimension count, but the supplied text does not contain the differential-rank argument needed to prove dominance; the central existence theorem is therefore conditional on a step that is not verifiable from the presented material.","major_comments":[{"comment":"The existence of rank-2 Ulrich bundles on the general double cover of P^3 branched in degree 2m=4,6,8 is reduced, via Proposition 3.56, to proving that the general branch polynomial b of degree 2m lies in the image of Φ(p_0,...,p_4)=p_0^2+p_1p_2+p_3p_4. The introduction says only that 'dimensional estimates on polynomials' show the possible cases; but a dimension inequality is necessary, not sufficient, for dominance. For m=4 the domain has dimension 5·C(7,3)=175 and the target C(11,3)=165, so a count cannot decide the question. The decisive check is whether the differential dΦ is surjective at a general point, i.e. whether the degree-8 part of the ideal (p_0,p_1,p_2,p_3,p_4) generated by five general quartics equals all of S_8. The text currently does not provide this check. Please state the genericity assertion as an explicit lemma in Section 3.2.4 and prove the differential-rank statement, or indicate clearly where in the manuscript it is proved.","section":"Introduction, 'Our work'; Theorem 3.45"},{"comment":"The claimed reduced moduli components of dimension 5, 6 and 0 depend on smoothness of the relevant Hilbert scheme of curves Y, which the introduction itself reduces to the same algebraic condition: whether the degree-2m part of the ideal generated by the p_i contains all degree-2m polynomials. For m=4, the octic double-solid case, this is exactly the surjectivity of dΦ discussed above. Without a proof of that surjectivity, the smoothness and dimension statements for the moduli space in the Calabi-Yau case are unsupported. The dimension-0 claim is especially delicate, since it requires showing that a general such bundle has no first-order deformations on the relevant component; a dimension count alone does not imply that.","section":"Theorem 3.45 and Section 4.1.1"},{"comment":"The polynomial decomposition b=p_0^2+p_1p_2+p_3p_4 is the single load-bearing assumption of the main existence theorem. The paper states that the general b admits such a presentation for m=2,3,4, but the presented text does not contain a complete proof. If the proof in Section 3.2.4 relies on a generic smoothness or open-orbit argument, it should be written out explicitly, including the computation of the codimension of the image or of the rank of dΦ. As written, a reader cannot distinguish between a true dominance theorem and an unproved assertion.","section":"Section 3.2.4 / Proposition 3.56"}],"minor_comments":[{"comment":"In the paragraph on morphisms to Grassmannians, 'symnolf' should read 'symbol'.","section":"Introduction"},{"comment":"The displayed theorem 'Lemma 3.63, Proposition 3.64' states a conclusion for n≥3 but then refers to a 'surjective contraction to P3'. For n>3 this should presumably be P^n, or the statement should be restricted to n=3; please clarify.","section":"Introduction, 'Our work'"},{"comment":"The phrase 'The last condition in this statement is equivalent to the last condition in the previous one' is confusing because the two statements use different formulas; please spell out the equivalence.","section":"Corollary 2.16"},{"comment":"Some references are cited with incomplete data, for example '[Bea][p. 18-20]' in the discussion of tentative counterexamples; please supply the full reference.","section":"Bibliography"}],"recommendation":"major_revision","confidential_remarks":"The paper reads as a PhD thesis with a substantial survey component; the genuinely new material is concentrated in Sections 3.2.4 and Chapter 4. The editor should ensure that the full proof of the polynomial genericity statement in Section 3.2.4 is available to referees; if it consists only of the dimension count mentioned in the introduction, the main theorem is not established. The structural equivalence in Propositions 3.54 and 3.56 is solid and likely publishable once the existence step is rigorously supplied."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nHere's my take on arXiv:2507.09345. This is a serious PhD thesis, not a throwaway preprint, and it deserves referee time. What's genuinely new: the divisorial-cover existence theorem via matrix factorizations (3.30), the if-and-only-if between existence of rank 2 Ulrich sheaves and the factorization b = p0^2 + p1 p2 + p3 p4 (3.54/3.56), and the applications—rank 2 Ulrich bundles on general double solids with branch degree 4, 6, 8, plus the octic double solid as a Calabi-Yau example with a moduli component of dimension 0. The Hartshorne-Serre treatment is careful, and the author is upfront about where he thinks a cited proof has a gap (Remark 4.30). That is honest work.\n\nThe soft spot is exactly where the stress-test lands: Theorem 3.45 reduces existence to showing that the general degree-2m branch polynomial b admits the representation b = p0^2 + p1 p2 + p3 p4. The introduction says 'dimensional estimates on polynomials' settle it, but that is necessary, not sufficient. In the m=4 case the source has dimension 175 and the target 165, so a crude dimension count leaves room for dominance but doesn't prove it. The real question is whether the differential of the map (p0,...,p4) -> p0^2+p1p2+p3p4 is surjective at a general point—equivalently, whether the degree-8 part of the ideal (p0,...,p4) generated by five general quartics is all of S_8. The full argument is supposed to be in Section 3.2.4, but that section isn't included in the text I received, so I cannot certify the central existence claim from what's in front of me. The moduli-space dimensions (5, 6, 0) and the smoothness claims rest on the same genericity, so this one step carries a lot of weight.\n\nI don't see an internal contradiction, and the structural theorem (3.56) is a genuine equivalence, not a tautology. But the paper's main claim for the octic double solid depends on a polynomial dominance statement that an independent referee must verify from Section 3.2.4.\n\nWho is this for: researchers working on Ulrich bundles, double covers, Fano and Calabi-Yau 3-folds. It deserves a serious referee rather than a desk rejection. I'd recommend sending it to peer review, with the explicit instruction that the polynomial factorization genericity in Section 3.2.4 be checked carefully.\n\nBest.","headline":"Serious thesis with genuinely new results on rank-2 Ulrich bundles on double solids, but the central existence theorem rests on a polynomial-factorization genericity step that the excerpt does not certify; referee should verify it.","tokens_in":65708,"tokens_out":3126,"would_cite":false,"duration_ms":35327,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14J60","14J30","14F05"],"pacs":[],"model":"deepseek-v4-flash","headline":"A matrix-factorisation construction and a Hartshorne-Serre construction prove that the general double cover of P^3 branched along a divisor of degree 4, 6, or 8 admits a stable rank 2 Ulrich bundle, making rank 2 the minimal possible.","keywords":["Ulrich bundles","double covers","double solids","Hartshorne-Serre correspondence","matrix factorisations","moduli spaces of sheaves","rank 2 bundles","weighted projective spaces"],"falsifier":"Fix m=2,3,4 and compute the dimension of the Zariski closure of the image of the map (p0,p1,p2,p3,p4)↦$p0^{2}$+p1p2+p3p4 on the space of 5-tuples of degree-m homogeneous polynomials in four variables. If for any one of these m the closure has dimension strictly smaller than dim $H^{0}$($P^{3}$,O(2m)), exhibit a smooth degree-2m branch divisor outside the image; such a divisor would give a smooth double solid with no rank 2 Ulrich sheaf, contradicting the theorem.","tokens_in":64558,"feed_emoji":"🧩","tokens_out":10097,"duration_ms":109558,"temperature":0.7,"pith_summary":"Fixed a polarised variety, an Ulrich sheaf is one whose twisted cohomology vanishes for all twists between -n and -1; existence and minimal rank are the questions. This thesis proves that every divisorial covering of P^n—a finite cover factoring through a divisor in the weighted projective space P($1^{{n+1}}$,m)—admits Ulrich sheaves, via matrix factorisations. For smooth double covers, the main theorem states that the general cover X→P^n admits a rank 2 Ulrich sheaf if and only if n=3 and the branch divisor has degree 4, 6, or 8; rank 2 is then minimal because the cyclic Picard group excludes Ulrich line bundles. The proof exhibits each such bundle in an exact sequence 0→O_X→E→I_Y(m)→0, where Y is a curve mapped isomorphically by the cover onto a complete intersection of two degree-m surfaces, and it produces reduced moduli components of dimension 5, 6, and 0 in the three cases. If correct, these double solids become among the few 3-folds for which the minimal Ulrich rank is known and sharp.","feed_headline":"General double solids of degrees 4, 6, 8 carry rank-2 Ulrich bundles","feed_subtitle":"A polynomial identity p0²+p1p2+p3p4 decides the minimal-rank sheaves on these threefolds.","key_machinery":"The load-bearing mechanism is the polynomial identity b=$p0^{2}$+p1p2+p3p4 together with the Hartshorne-Serre correspondence, the rank-2 dictionary between vector bundles and codimension-2 subvarieties expressed by extensions 0→O_X→E→I_Y(m)→0. Geometrically, the identity is read as choosing a complete intersection {p1=p3=0} on which the branch locus restricts to a square, so the curve lifts to the double cover and generates the required extension; algebraically, it is the existence of this presentation for the general degree-2m branch polynomial that makes the general double solid admit the curve Y. The cohomological vanishing defining Ulrich sheaves is exactly what forces Y to have the right no-intermediate-cohomology profile, and a matrix-factorisation construction for divisorial coverings supplies the broader existence theorem for Ulrich sheaves on X⊂P($1^{{n+1}}$,m).","core_discovery":"The central claim is an equivalence for a smooth double cover f:X→$P^{3}$ branched along a degree-2m divisor with equation b=0 and Pic(X)=ZH: the branch polynomial admits a presentation b=$p0^{2}$+p1p2+p3p4 with homogeneous p_i of degree m if and only if X carries a rank 2 Ulrich sheaf E, and this happens exactly when a curve Y⊂X exists that is mapped isomorphically by f onto a complete intersection of two degree-m hypersurfaces; the bundle then fits 0→O_X→E→I_Y(m)→0. The paper proves by a genericity argument, deferred to Section 3.2.4, that for m=2,3,4 the general branch polynomial has such a presentation, so the general double solid carries a stable rank 2 Ulrich bundle. In those cases the Hilbert scheme of the curves Y yields generically smooth moduli components of the expected dimension (5, 6, 0), and for m=2,3 the same extension-and-deformation machinery produces stable Ulrich bundles of every admissible rank. For m=4 the rank 2 bundles are spherical, and the same characterization rules out rank 2 Ulrich sheaves on all other double covers of P^n with n≥3.","pith_inferences":["Beyond the paper: the polynomial-presentation condition b=p0^2+p1p2+p3p4 could be tested for degree-2m branch loci with m≥5, where the parameter count already suggests failure of dominance; a negative answer would make the octic double solid the sharp cutoff case among double solids.","Beyond the paper: the curves Y appearing as zero loci give a direct geometric handle on the other extremal contraction of P(E); one could check whether the Abel-Jacobi map to the intermediate Jacobian is generically finite for the sextic case, as the paper recalls it is for the quartic case.","Beyond the paper: dominance of the map (p0,p1,p2,p3,p4)↦p0^2+p1p2+p3p4 can be verified computationally over a finite field of large characteristic for m=2,3,4; dominance there would imply the characteristic-zero genericity statement by spreading out, giving an independent check of the existence half.","Beyond the paper: the Hartshorne-Serre description suggests a testable extension to cyclic covers of degree d>2, where a rank-2 Ulrich sheaf should correspond to expressing the branch data as a sum of two d-th power forms, a problem with the same parameter-count flavour."],"forward_implications":["Every divisorial covering of P^n admits an Ulrich sheaf over any field, giving an upper bound on the minimal Ulrich rank attached to the defining equation.","On smooth double covers of P^3, the minimal Ulrich rank is 2 precisely for branch degrees 4, 6, and 8; outside this range, and for n≠3, no rank 2 Ulrich sheaf exists.","The moduli spaces of rank 2 Ulrich bundles on the general quartic, sextic, and octic double solids contain reduced components of dimension 5, 6, and 0 respectively, and the bundles are slope-stable.","Stable Ulrich bundles of every rank r≥2 exist on the general quartic double solid, with generically smooth moduli components of dimension r^2+1; on the general sextic double solid, every even rank 2ρ occurs, with components of dimension 5ρ^2+1.","On a general quartic double solid, the restriction map from Ulrich bundles on X to Ulrich bundles on a smooth hyperplane section is generically étale and the target is irreducible, yielding the stated interpolation result for curves through prescribed points."],"supporting_citations":[{"why":"Establishes the correspondence between Cohen-Macaulay modules on hypersurface rings and matrix factorisations, which underlies the algebraic construction of Ulrich sheaves.","marker":"[Eis80]"},{"why":"Proves existence of matrix factorisations under general assumptions on the ring, yielding Ulrich modules on complete intersections and supplying the key input for Theorem 3.30.","marker":"[HUB91]"},{"why":"Shows existence of Ulrich sheaves on double covers of projective space in the degree-2 case, the starting point extended here to divisorial covers and rank 2 on double solids.","marker":"[MNP25]"},{"why":"Treats Ulrich sheaves on cyclic coverings of arbitrary degree, giving the context and comparison for the double-cover results.","marker":"[PP24]"},{"why":"Provides the original Hartshorne-Serre correspondence for rank 2 bundles, used to construct E from the curve Y.","marker":"[Har78]"},{"why":"Extends the correspondence to reflexive sheaves and codimension-2 subvarieties, backing the general version reviewed in Chapter 2.","marker":"[Har80]"},{"why":"Supplies the extension-and-deformation strategy for constructing higher-rank stable Ulrich bundles from low-rank ones.","marker":"[CFK23b]"},{"why":"Introduces the sheaf-theoretic definition of Ulrich sheaves and the linear-resolution characterization on which all existence and vanishing arguments rest.","marker":"[ES03]"}],"fun_headline_variants":["Degree 4,6,8 double solids: exactly where rank-2 Ulrich bundles live","Rank-2 Ulrich sheaves exist only on double solids of degree 4,6,8","Double solids of degree 4,6,8 admit rank-2 Ulrich sheaves","A polynomial identity decides which double solids carry Ulrich rank 2","For double solids, Ulrich rank 2 occurs exactly at degrees 4,6,8"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The existence claims for the general double solids reduce to the statement that a general homogeneous polynomial of degree 2m=4,6,8 in four variables can be written as $p0^{2}$+p1p2+p3p4 with each p_i of degree m; if that genericity statement fails for any of the three degrees, the corresponding rank 2 existence claim for the general double cover fails.","fun_headline_variants_meta":{"raw":{"variants":["Degree 4,6,8 double solids: exactly where rank-2 Ulrich bundles live","Rank-2 Ulrich sheaves exist only on double solids of degree 4,6,8","Double solids of degree 4,6,8 admit rank-2 Ulrich sheaves","A polynomial identity decides which double solids carry Ulrich rank 2","For double solids, Ulrich rank 2 occurs exactly at degrees 4,6,8"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000682,"raw_usage":{"total_tokens":3133,"prompt_tokens":1020,"completion_tokens":2113,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":636,"completion_tokens_details":{"reasoning_tokens":2017}},"tokens_in":636,"tokens_out":2113,"duration_ms":18460,"temperature":1.0,"reasoning_tokens":2017,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:57:49.834039+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fix m=2,3,4 and compute the dimension of the Zariski closure of the image of the map (p0,p1,p2,p3,p4)↦$p0^{2}$+p1p2+p3p4 on the space of 5-tuples of degree-m homogeneous polynomials in four variables. If for any one of these m the closure has dimension strictly smaller than dim $H^{0}$($P^{3}$,O(2m)), exhibit a smooth degree-2m branch divisor outside the image; such a divisor would give a smooth double solid with no rank 2 Ulrich sheaf, contradicting the theorem.","supporting_citations":[],"review_version":1}