{"id":"2bd6ad12-d892-4af4-8aa9-e958a780ccca","arxiv_id":"1908.10986","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For general quartic double solids, two varieties are isomorphic if and only if their Kuznetsov components are equivalent, without assuming the equivalence has Fourier-Mumford type.","lead":"Mathematicians have shown that a class of three-dimensional shapes called quartic double solids is completely determined by the internal structure of its derived category, a kind of algebraic shadow. This gives a 'categorical Torelli theorem' without an extra technical condition required by earlier proofs.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 5.8 is the load-bearing gap: the non-Fourier-Mukai family construction is cited to [BMMS12, Lemma 5.2] rather than proven, and without it Theorem 5.1 does not go beyond the previously known Fourier-Mukai case.","rationale":"The paper's main new claim is the refined categorical Torelli theorem for general quartic double solids, and its proof is concentrated in Section 5. The moduli-space description in Section 2–4 appears carefully developed and internally consistent: the wall-crossing argument is explicit, and the low-degree computations for d=2 and d=1 are detailed. I found no independent reason to doubt those parts. The decisive point is the passage from a bijection of closed points, obtained in Proposition 5.7, to a genuine morphism Y' → M for an arbitrary equivalence u. That passage is exactly the content of Section 5.2, and it rests entirely on Lemma 5.8, whose proof is omitted and merely referenced. Since the paper explicitly says the lemma 'follows from the same argument' and gives no verification that the hypotheses of [BMMS12, Lemma 5.2] hold for quartic double solids, this is an identifiable gap rather than a disagreement with consensus. The reader's verdict identifies the same weakest assumption. I therefore recommend keeping the conditional verdict: the Torelli theorem is plausible and the moduli-space results are substantial, but the unconditional statement for non-Fourier-Mukai equivalences is not fully supported until Lemma 5.8 is proved in this setting.","tokens_in":36692,"tokens_out":6603,"duration_ms":70437,"concrete_test":"Write out an independent proof of Lemma 5.8 for Y' a general quartic double solid, adapting [BMMS12, Lemma 5.2] line by line and replacing every cubic-threefold input with the corresponding facts for Y' (cohomology of O_Y'(-N) and O_Y'(1)(-N), the Serre functor of Ku(Y'), and the numerical lattice (3)). Then verify directly that the resulting object E~ satisfies i_s^*E~ ≃ F((i'_s)_*E'_s) for every closed point s ∈ Y', as Lemma 5.9 claims. If the proof cannot be completed, Theorem 5.1 should be restated with the extra hypothesis that u is of Fourier-Mukai type.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 5.1 requires a morphism Y' → M whose points are the images u(E'_s). For an equivalence u that is not of Fourier-Mukai type, the family E~ in §5.2 is constructed only through Lemma 5.8. That lemma states that the complex F•_m admits a unique split right convolution Gm ≃ Hm ⊕ Em with specified cohomological amplitude, but its proof is not given: the text says only 'The following lemma follows from the same argument given in [BMMS12]' and then 'Proof. See [BMMS12, Lemma 5.2].' This is not a routine application, because the setting here is a quartic double solid rather than the cubic threefold of [BMMS12]; the terms F(O_Y'(-N_i)) ⊠ L^{⊗-r_i} are objects of Db(Y×Y'), generally complexes rather than sheaves, and the asserted uniqueness of the right convolution depends on nontrivial Hom-vanishing and amplitude checks. Lemma 5.9 then relies on the same uniqueness assertion to identify the restriction i_s^*E~ with F((i'_s)_*E'_s). Without a proof of Lemma 5.8, Proposition 5.7 supplies only a bijection on closed points, not a morphism of moduli spaces, and the rational-connectedness argument in §5.3 cannot be run. If Lemma 5.8 is false or requires extra hypotheses, Theorem 5.1 is not established for arbitrary equivalences, and the improvement over [BT16, Cor. 3.1(iii)] is lost.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies moduli spaces of stable objects of a fixed numerical class w in the Kuznetsov component Ku(Y) of a Fano threefold Y of index 2 and Picard rank 1. Using the stability condition of Bayer–Lahoz–Macrì–Stellari and wall-crossing in tilt-stability, the authors describe the moduli space M_σ(w) for all degrees d = 1,...,5: it contains a component isomorphic to Y, and for d ≤ 2 a second component parametrizing roots on hyperplane sections. For d = 2 (quartic double solids) and d = 1 (double Veronese cones) they study the Abel–Jacobi map to the intermediate Jacobian. The main application is a categorical Torelli theorem (Theorem 5.1): for two general quartic double solids Y and Y', any equivalence Ku(Y') ≅ Ku(Y) implies Y' ≅ Y, without the Fourier–Mukai assumption that was needed in [BT16]. The proof constructs, from such an equivalence, a morphism Y' → M_σ(w) via a universal family built by convolutions, and identifies its image with the component Y.","tokens_in":37064,"tokens_out":3188,"duration_ms":31511,"significance":"If fully established, Theorem 5.1 is a substantial result: it would make the Kuznetsov component a complete categorical invariant for general quartic double solids, removing the Fourier–Mukai hypothesis from the earlier Bernardara–Tabuada result. The paper also gives a fairly complete description of M_σ(w) in all degrees, with explicit calculations of Ext spaces, Chern classes, and wall-crossing behavior. The low-degree descriptions (d = 1, 2) are supported by concrete geometric arguments, and the Abel–Jacobi analysis for quartic double solids contains interesting new information about the intersection of the two components. The paper is well structured and careful in most of its technical sections. However, the main Torelli theorem depends on Lemma 5.8, which is asserted by citation to [BMMS12] rather than proved; this is a load-bearing gap for the non-Fourier–Mukai case, and without it the claimed improvement over [BT16] is not established.","major_comments":[{"comment":"Lemma 5.8 is the crucial step that allows the construction of the family E~ for an equivalence u that is not of Fourier–Mukai type. The lemma asserts that the complex F•_m of (17) admits a unique split right convolution Gm ≅ Hm ⊕ Em with specific cohomological amplitude, but its proof is only a citation to [BMMS12, Lemma 5.2]. This is not a routine adaptation: in the present setting the terms F(O_Y'(-N_i)) ⊠ L^{⊗-r_i} are objects of D^b(Y×Y') and are generally complexes, not sheaves, and the uniqueness of the right convolution requires explicit Hom-vanishing and amplitude checks that depend on the geometry of quartic double solids. Because Proposition 5.7 supplies only a bijection on closed points, and Lemma 5.9 and the morphism α in §5.3 rely on this same uniqueness, the proof of Theorem 5.1 for arbitrary equivalences is incomplete. The authors should either provide a full proof of Lemma 5.8 or a precise reduction to the cubic-threefold argument that verifies each hypothesis needed from [BMMS12].","section":"§5.2, Lemma 5.8"},{"comment":"Lemma 5.9 asserts that i_s^*(E~) ≅ F((i'_s)_*E'_s) for every closed point s ∈ Y', arguing that the two sides are both right convolutions of the restricted complex i_s^*(F•_m) and that this convolution is unique 'by the same argument as in Lemma 5.8'. Since Lemma 5.8 is not proved, the uniqueness of the restricted convolution is also unsupported. This is not an independent technicality: without Lemma 5.9 the family E~ cannot be shown to have the correct fiberwise objects, and hence α cannot be identified with the pointwise assignment u(E'_s). The dependence of Lemma 5.9 on Lemma 5.8 should be made explicit and the missing argument supplied.","section":"§5.3, Lemma 5.9"},{"comment":"The conclusion that the morphism α: Y' → M factors through the component Y and is an isomorphism uses the claim that 'α is, in particular, a morphism dominating one of the components of M and α is birational onto its image.' This claim is not justified in the text. The bijection on closed points from Proposition 5.7, together with the construction of α, may imply surjectivity onto a component, but birationality and the eventual isomorphism require a separate argument (for instance, that the family E~ is generically an isomorphism or that the induced map on tangent spaces is an isomorphism at a general point). Since the preceding lemmas are the only place where the fiberwise identifications are established, this step also inherits the gap in Lemma 5.8. The authors should spell out the birationality argument or replace it with a direct verification that α is an isomorphism using the bijection on closed points and smoothness.","section":"§5.3, proof of Theorem 5.1"}],"minor_comments":[{"comment":"The text 'In this section, . We start with a series of lemmas' contains a dangling period and an incomplete sentence; it should be revised.","section":"§5.1, first paragraph"},{"comment":"The symbol Y is used both for the Fano threefold and for the locus in M_σ(w) isomorphic to Y. This is a common but potentially confusing overload; a different notation (e.g., Y_M or Y_0) for the moduli component would improve readability.","section":"§2.2, Proposition 2.9"},{"comment":"In the resolution (15), the notation O_{Y'}(-N_i) ⊠ (L^{⊗-r_i}) should specify that L is an ample line bundle on Y' (presumably O_{Y'}(1)) and justify the existence of such a resolution with the stated properties; currently the choice of L and the vanishing conditions are left implicit.","section":"§5.2, Eq. (15)"},{"comment":"The computation of χ(u(E),u(E)) = -(a+b)^2 - b^2 for [u(E)] = av + bw is correct but the sign convention for the Euler form should be stated explicitly, since the intersection matrix in §1 has the opposite sign convention in some places.","section":"§5.1, Lemma 5.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is promising and the main geometric descriptions appear sound, but the central categorical Torelli theorem currently hinges on an unproved lemma that is only cited from [BMMS12]. This is a fixable gap if the authors can supply the missing proof, but as it stands the result does not go beyond the previously known Fourier–Mukai case. The novelty of the paper—and the strength of the claim in Theorem 5.1—depends entirely on this point, so I would ask the editor to insist on a complete proof of Lemma 5.8 (and Lemma 5.9) before acceptance. The rest of the paper is generally careful and original, and the low-degree results are worth publishing once the Torelli proof is completed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a solid paper with one load-bearing gap. The genuinely new content is the low-degree description of M_sigma(w) for d = 1 and d = 2, plus the attempt to remove the Fourier-Mukai assumption from the Bernardara-Tabuada Torelli result for quartic double solids. Sections 2 through 4 are careful: explicit Ext computations, Chern-class checks, wall-crossing analysis, and Abel-Jacobi maps all hang together. Theorem 4.4 and Theorem 4.11 look right to me, and the intersection statements involving the ramification locus and the curve C are plausible and supported. Independent of the Torelli application, this moduli-space part is a real contribution.\n\nThe soft spot is exactly where the stress-test lands: Lemma 5.8. The lemma asserts existence and uniqueness of a split right convolution for the complex F_m^bull, and the proof is simply \"See [BMMS12, Lemma 5.2].\" That is not a routine application. The setting here is a quartic double solid, not the cubic threefold of BMMS12; the terms F(O_Y'(-N_i)) are objects of D^b(Y x Y'), generally complexes rather than sheaves, and the claimed uniqueness of the convolution requires genuine Hom-vanishing and amplitude checks that are not supplied. Without Lemma 5.8, Proposition 5.7 gives only a bijection on closed points; you cannot promote it to a morphism Y' -> M, and the rational-connectedness argument in Section 5.3 cannot run. So Theorem 5.1, as stated, is not fully established. This is not a manufactured objection: the manuscript itself signals the gap by relegating the proof to a reference. If Lemma 5.8 is false or needs extra hypotheses, the improvement over [BT16] evaporates.\n\nThe rest of the paper does not depend on that lemma. I saw no circularity: the moduli-space description used in the Torelli argument is proved independently in Sections 2-4. The citation pattern looked fair; Section 3 is recapitulation, but it is clearly labeled as such. The paper is for people working on Bridgeland stability, moduli of objects in Kuznetsov components, and categorical Torelli questions for Fano threefolds.\n\nRecommendation: send it to peer review. A serious referee should be asked to produce a complete proof of Lemma 5.8, or to state explicitly which additional hypotheses make the BMMS12 argument carry through. Until that is done, Theorem 5.1 should be treated as conditional.","headline":"Strong moduli-space work that deserves refereeing; the advertised Torelli theorem is conditional on a missing proof of the non-Fourier-Mukai family construction in Lemma 5.8.","tokens_in":37616,"tokens_out":1961,"would_cite":true,"duration_ms":21042,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14F08","14J45","14D20"],"pacs":[],"model":"deepseek-v4-flash","headline":"For general quartic double solids, an equivalence of Kuznetsov components forces the two threefolds to be isomorphic.","keywords":["derived categories","Bridgeland stability conditions","Fano threefolds","moduli spaces","Abel-Jacobi map","categorical Torelli theorem","quartic double solids","Kuznetsov component"],"falsifier":"Take a general quartic double solid, choose the resolution (15) used in Section 5.2 for some $m$, and compute whether the complex $F_m^\\bullet$ admits a unique split right convolution $G_m\\simeq H_m\\oplus E_m$ with the cohomological support stated in Lemma 5.8. If for any $m$ two non-isomorphic right convolutions exist, or if the required vanishing fails, the family $\\widetilde E$ used to produce the morphism $Y'\\to\\mathcal M_\\sigma(w)$ is not well defined and the proof of the categorical Torelli theorem collapses.","tokens_in":36483,"feed_emoji":"🧩","tokens_out":15286,"duration_ms":135071,"temperature":0.7,"pith_summary":"This paper establishes that a general quartic double solid—a double cover of $\\mathbb P^3$ branched over a quartic K3 surface—is determined up to isomorphism by its Kuznetsov component, the nontrivial part of a semiorthogonal decomposition of its derived category of coherent sheaves. If two such threefolds have equivalent Kuznetsov components, they are isomorphic as varieties, with no requirement that the equivalence be of Fourier–Mukai type. The proof works by studying the moduli space of stable objects of one numerical class $w$ in the Kuznetsov component, showing it has two three-dimensional irreducible components: one copy of the threefold itself and one component whose points correspond to roots on hyperplane sections. An Abel–Jacobi map distinguishes the two components geometrically, and an arbitrary equivalence is shown to respect this geometry. If correct, this is a categorical Torelli theorem of the strongest expected form for these threefolds.","feed_headline":"Kuznetsov component pins down quartic double solids","feed_subtitle":"For these threefolds, an equivalence of Kuznetsov components forces an isomorphism of varieties, with no Fourier–Mukai assumption.","key_machinery":"The central object is the moduli space $\\mathcal M_\\sigma(w)$ of semistable objects of numerical class $w$ in the Kuznetsov component $\\operatorname{Ku}(Y)$, where $\\sigma$ is the Bridgeland stability condition on $\\operatorname{Ku}(Y)$ induced from weak stability conditions on $D^b(Y)$ by the general criterion of [BLMS17]. The class $w$ is the vector $w=H-\\tfrac12H^2+(\\tfrac16-\\tfrac1d)H^3$ in the numerical Grothendieck group, and the objects $E_p$ of that class form the subvariety $\\mathcal Y\\simeq Y$. Three mechanisms carry the argument: (1) wall-crossing in the $(\\alpha,\\beta)$-plane identifies $\\mathcal M_\\sigma(w)$ with a moduli space of $\\sigma_{\\alpha,-1/2}$-semistable complexes below the unique wall and compares it with the Gieseker moduli space $\\mathcal M_G(w)$; (2) the Abel–Jacobi map $F\\mapsto\\Phi(c_2(F))$ to the intermediate Jacobian $J(Y)$ separates the two components for $d\\le 2$; (3) for the Torelli theorem, a convolution construction assembles a universal family for the objects $u(E'_s)$ without assuming $u$ is a Fourier–Mukai functor.","core_discovery":"The central discovery is that for a general quartic double solid $Y$, the Kuznetsov component $\\operatorname{Ku}(Y)$ is a complete categorical invariant: given another general quartic double solid $Y'$, any equivalence of triangulated categories $\\operatorname{Ku}(Y')\\simeq\\operatorname{Ku}(Y)$ forces an isomorphism $Y'\\simeq Y$. The engine is the moduli space $\\mathcal M_\\sigma(w)$ of $\\sigma$-stable objects of class $w$ in $\\operatorname{Ku}(Y)$, where $w=H-\\tfrac12H^2+(\\tfrac16-\\tfrac1d)H^3$; for $d=2$ this moduli space has two three-dimensional irreducible components, one isomorphic to $Y$ and parametrizing objects $E_p$ defined by the triangle $\\mathcal O(-1)[1]\\to E_p\\to I_p\\to \\mathcal O(-1)[2]$, the other parametrizing roots of hyperplane sections. The components meet exactly along the ramification locus $R$, and the Abel–Jacobi map $\\Psi(F)=\\Phi(c_2(F))$ contracts the $Y$-component to a point while being a generic embedding on the other component. The proof of the Torelli statement shows that any equivalence sends the stable objects of class $w$ to stable objects of class $w$ up to shift, uses a convolution construction to turn the induced bijection on closed points into a morphism $Y'\\to\\mathcal M_\\sigma(w)$, and then uses rational connectedness to force the image into the component isomorphic to $Y$.","pith_inferences":["One could test whether the convolution lemma used in the proof can be proved in general; if so, the same strategy should give a categorical Torelli theorem for the degree-1 case (the double Veronese cone) once the homological-dimension obstruction (the heart has homological dimension 3 there) is handled.","Because the intersection $\\mathcal Y\\cap\\mathcal C$ is the ramification K3 surface, the abstract moduli space $\\mathcal M_\\sigma(w)$ may encode enough information to reconstruct the branch quartic, making the Torelli statement constructive rather than existence-only.","The numerical observation that an equivalence may send class $w$ to $2v-w$, corrected by a rotation functor, suggests a closer relation between autoequivalences of $\\operatorname{Ku}(Y)$ and the wall-crossing group of the moduli space; testing whether the rotation matches the reflection across the wall could yield a shorter proof of the shift-invariance step."],"forward_implications":["For general quartic double solids, $\\operatorname{Ku}(Y)$ is a complete categorical invariant: $Y'\\simeq Y$ if and only if $\\operatorname{Ku}(Y')\\simeq\\operatorname{Ku}(Y)$, with no Fourier–Mukai hypothesis on the equivalence.","Consequently any equivalence $\\operatorname{Ku}(Y')\\simeq\\operatorname{Ku}(Y)$ of general quartic double solids implies a Fourier–Mukai equivalence between the two categories, by combining Theorem 5.1 with the earlier result [BT16, Prop. 3.5].","The moduli space $\\mathcal M_\\sigma(w)$ for a general quartic double solid has exactly two irreducible components, $\\mathcal Y\\simeq Y$ and $\\mathcal C$, smooth outside their intersection at the ramification locus $R$; the Abel–Jacobi map contracts $\\mathcal Y$ to a point and embeds $\\mathcal C$ generically.","For degree $d\\le 2$ the same two-component description holds, while for $d\\ge 3$ the moduli space is irreducible and projective; in degree $1$ the second component is the Fano surface of lines, meeting $\\mathcal Y$ along the curve $C$."],"supporting_citations":[{"why":"Supplies the stability condition on the Kuznetsov component that defines the moduli space $\\mathcal M_\\sigma(w)$ and controls which objects are semistable.","marker":"[BLMS17]"},{"why":"Provides the convolution technique (Lemma 5.2) that the paper adapts to build the universal family from an arbitrary equivalence; also the cubic-threefold categorical Torelli model.","marker":"[BMMS12]"},{"why":"Proves the categorical Torelli statement under a Fourier–Mukai assumption; Theorem 5.1 removes that assumption, and its Proposition 3.5 gives the Fourier–Mukai-equivalence corollary.","marker":"[BT16]"},{"why":"Supplies the Abel–Jacobi/TBS differential technique and the generality assumptions on the branch quartic used to control hyperplane sections and the map's rank.","marker":"[Wel81]"},{"why":"Gives base change for semiorthogonal decompositions, used to make the universal family construction in Lemma 2.8 and Proposition 2.9 compatible with families.","marker":"[Kuz11]"},{"why":"Computes the Serre functor of the Kuznetsov component and defines the rotation functor $R$, used in Lemma 5.6 to adjust the numerical class of objects under an equivalence.","marker":"[Kuz15]"},{"why":"Describes the wall and chamber structure for these Fano threefolds (walls are concentric semicircles), used to locate the single wall separating the Gieseker and Bridgeland moduli spaces.","marker":"[BMSZ17]"},{"why":"Provides the classification of roots and lines on del Pezzo surfaces (126 roots, 56 lines) used to describe the component $\\mathcal C$ and to relate roots to conics.","marker":"[Dol12]"}],"fun_headline_variants":["Categorical Torelli theorem for quartic double solids","Kuznetsov moduli spaces distinguish quartic double solids","Stable object moduli yield a Torelli result","Moduli on Kuznetsov components recover the threefold","Complete invariant from Kuznetsov component moduli"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument that the constructed family of objects really gives a morphism between the two threefolds depends on an unproved technical claim about decomposing a certain complex in a canonical way; without that claim, the equivalence is only known to induce a bijection between sets of closed points, which is not enough to conclude the two threefolds are isomorphic.","fun_headline_variants_meta":{"raw":{"variants":["Categorical Torelli theorem for quartic double solids","Kuznetsov moduli spaces distinguish quartic double solids","Stable object moduli yield a Torelli result","Moduli on Kuznetsov components recover the threefold","Complete invariant from Kuznetsov component moduli"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00022,"raw_usage":{"total_tokens":1467,"prompt_tokens":988,"completion_tokens":479,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":604,"completion_tokens_details":{"reasoning_tokens":399}},"tokens_in":604,"tokens_out":479,"duration_ms":5433,"temperature":1.0,"reasoning_tokens":399,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:28:20.919113+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a general quartic double solid, choose the resolution (15) used in Section 5.2 for some $m$, and compute whether the complex $F_m^\\bullet$ admits a unique split right convolution $G_m\\simeq H_m\\oplus E_m$ with the cohomological support stated in Lemma 5.8. If for any $m$ two non-isomorphic right convolutions exist, or if the required vanishing fails, the family $\\widetilde E$ used to produce the morphism $Y'\\to\\mathcal M_\\sigma(w)$ is not well defined and the proof of the categorical Torelli theorem collapses.","supporting_citations":[],"review_version":1}