{"id":"6e93a85f-1e85-4c82-a0e2-5cf0120c9ae1","arxiv_id":"2608.07853","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For general Gushel-Mukai threefolds, Bridgeland moduli spaces on the Kuznetsov component are normal, and for primitive numerical classes of odd square they are irreducible.","lead":"A mathematics paper proves that certain moduli spaces of objects in the derived category of Gushel-Mukai threefolds are normal, and that under a parity condition they are irreducible for general threefolds. A generalist should care because these spaces are central objects in the study of Fano threefolds and stability conditions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 6.1 is unproved as written: its proof asserts that the generic fibre Y=M_eta dominates C, which is false; since Theorem 6.3 uses Lemma 6.1 to propagate irreducibility from the special fibre, the main theorem lacks support until a correct proof is supplied.","rationale":"The reader's stated weakest assumption is the Hodge signature input in Lemma 4.7, but I find a more immediate and internally visible gap in Lemma 6.1, which the reader also flagged in the rationale. Lemma 6.1 is used in the proof of Theorem 6.3 exactly to pass from irreducibility of the special fibre to irreducibility over an open neighbourhood, and then to a general GM threefold. The proof of that lemma contains a false claim about the image of the generic fibre, so as written the deformation step is unsupported. The Hodge-signature question in Lemma 4.7 is a legitimate external-input concern, but it is a matter of checking a cited theorem from [BP23] and [Per19], whereas Lemma 6.1 is internally incoherent in the manuscript itself. I do not claim the main theorem is false; the lemma may be true and repairable, and the rest of the argument appears coherent. For that reason the verdict should remain CONDITIONAL: the paper should be accepted only after a correct proof of Lemma 6.1 (or a replacement argument) is supplied. My agreement with the reader is only partial because I would elevate the deformation lemma to the primary load-bearing concern, while the reader's weakest-assumption field points to the Hodge structure input.","tokens_in":20911,"tokens_out":21196,"duration_ms":201012,"concrete_test":"Settle Lemma 6.1 by attempting the following proof. Let eta be the generic point of C, let Y:=M_eta, and let Z be an irreducible component of M with pi(Z)=a. Consider the closure \\overline{Y} of the generic fibre in M; because pi is proper, \\overline{Y} is proper over C and its special fibre is a nonempty closed subscheme of the fibre M_a. Since M_a is irreducible and every fibre is reduced, this closed subscheme must be set-theoretically all of M_a. This forces Z to be contained in \\overline{Y}, contradicting the reducedness of M_a unless Z is not an irreducible component of M. If this argument can be completed, the lemma is true and only the written proof needs repair; if a counterexample with reduced normal fibres and irreducible central fibre but reducible nearby fibres can be constructed, then the deformation step of Theorem 6.3 collapses.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The most immediately load-bearing defect is in the deformation lemma 6.1, which is the step that carries irreducibility from the singular special fibre to nearby fibres and hence to a general Gushel--Mukai threefold. In the proof, after defining Y:=M_eta as the fibre over the generic point, the text says: 'As pi|_Y is proper and contains eta in its image, pi|_Y dominates C. Thus, pi|_Y is flat and therefore its fibres are of equidimension d.' This sentence is internally inconsistent: Y is the fibre over eta, so pi(Y) = {eta}, not C, and Y is not proper over C. The subsequent conclusion that Y intersects the contracted component Z, and that this contradicts the assumption on Z, is therefore unsupported. The lemma may well be true under the stated hypotheses -- properness plus reducedness and normality of all fibres may force the absence of components of M contracted to a point, because the closure of the generic fibre is proper over C and its special fibre is a closed reduced subscheme of the irreducible fibre M_a. But that argument is not what the paper gives. Since Theorem 6.3 explicitly invokes Lemma 6.1 to obtain irreducibility of fibres in a neighbourhood of the special fibre, the central claim is not established as written until Lemma 6.1 is either proved correctly or replaced by a valid argument.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the geometry of Bridgeland moduli spaces M_σ(Ku(X),v) attached to the Kuznetsov component Ku(X) of a smooth complex Gushel–Mukai threefold X. Theorem 1.2 states that if v_0 is primitive with v_0^2 ≥ 5, then for every m ≥ 1 the moduli space M_σ(Ku(X),m v_0) is normal. The main theorem, Theorem 1.1, states that for a primitive odd-square class v_0, the moduli space M_σ(Ku(X),v_0) is irreducible for a general GM threefold X. The proof of Theorem 1.2 uses the formality theorem of [CPZ24], a Serre-twisted quadratic Kuranishi model, a Hodge-theoretic rank estimate for τ-fixed stable objects via the CY2 cover, and an induction on dimension vectors to prove that the local zero fibre satisfies Serre's conditions. Theorem 1.1 follows from a Kaledin–Lehn–Sorger type blow-up Chern-class argument for connectedness of singular normal moduli spaces, and from a deformation argument that constructs a singular irreducible special fibre and propagates irreducibility to the general fibre. The paper is organized into sections on Kuznetsov components, stability conditions, normality, connectedness, and deformation.","tokens_in":21187,"tokens_out":27476,"duration_ms":259156,"significance":"If the main theorems are correct, they would be a substantial advance: they supply global irreducibility for moduli spaces of semistable objects in the Enriques category Ku(X), including classes with singular stable points, and they treat nonprimitive classes in the normality theorem. The approach is largely a derivation from published inputs — the formality statement [CPZ24], the Hodge-theoretic description of the CY2 cover [BP23, Per19], and the moduli space construction [PPZ23, BLM+21a] — with explicit parameter-free computations of Euler forms, dimension vectors, and Chern classes. The principal weakness is the deformation step: Lemma 6.1, which carries irreducibility from a special fibre to neighbouring fibres, is not proved correctly, and Theorem 6.3 depends on it. For this reason the main theorem is not established as written, although the gap appears to be localized and potentially fixable.","major_comments":[{"comment":"The proof of Lemma 6.1 contains a load-bearing error. After defining Y := M_η, the text states 'As π|_Y is proper and contains η in its image, π|_Y dominates C.' This is false: Y is the fibre over η, so π(Y) = {η}; π|_Y is not a proper morphism over C and does not dominate C. The subsequent conclusions that π|_Y is flat and that Y meets the contracted component Z are therefore unsupported. Since Theorem 6.3 invokes Lemma 6.1 to pass from the irreducible special fibre M_0 to irreducibility on an open neighbourhood of the base, the main theorem is not proved as written. The lemma may be true under the stated hypotheses, but a correct proof — using properness and separatedness of π and normality of the fibres, and not the false domination claim — is required.","section":"6.1, Lemma 6.1"},{"comment":"The reduction to Lemma 6.1 is not fully justified. Lemma 6.1 is stated for schemes, while p is a proper algebraic space; and in the case t_0 ∉ S^fl, the base-changed family over a curve is not flat at c_0, so the flatness assertion used in the proof of Lemma 6.1 is unavailable. The subsequent appeal to [Sta, Tag 0E1E] for local constancy of geometric connected components applies only over the flat locus S^fl. The proof should either establish an algebraic-space version of Lemma 6.1 that does not require flatness, or provide a different argument covering the non-flat neighbourhood of the special point.","section":"6.2, Theorem 6.3"}],"minor_comments":[{"comment":"The proof refers to 'Lemma 4.9' for the two equivariant lifts of S; the correct reference is Lemma 4.4.","section":"4.3, Lemma 4.7"},{"comment":"The phrase 'smaller then' should read 'smaller than'; for clarity, the proof should state explicitly that at a smooth point the actual dimension of M is ext^1(E,E), while the expected dimension is v_0^2+1, so the two numbers cannot coincide when E is τ-fixed.","section":"5.1, Lemma 5.1"},{"comment":"There are several wording errors: 'has one no analogous' should be 'there is no analogous' (Section 1.3), and the heading 'An Deformation Lemma' should be 'A deformation lemma' (Section 6.1).","section":"1.3 and 6.1"},{"comment":"Definition 3.1 is not self-contained: it refers to σ-semistable objects before σ has been defined and omits the standard slicing axioms (P(φ+1)=P(φ)[1] and existence and uniqueness of Harder–Nasem filtration). Please either give the full definition or replace it with a reference to [Bri07].","section":"3.1, Definition 3.1"},{"comment":"The Hodge-theoretic input from [BP23, Theorem 4.15(3)] and [Per19, Section 3.1] is load-bearing; please state precisely which result supplies the signature (2,20) of the real (1,1)-part of the Mukai Hodge structure on the CY2 cover, and confirm that it covers both ordinary and special GM threefolds.","section":"4.3, Lemma 4.7"},{"comment":"The assertion that primitivity of v_{0,s} together with Remark 3.4 implies stability coincides with semistability should be justified in one sentence: injectivity of the central charge forces every stable factor's class to lie on the same ray as v_{0,s}, hence to be an integral multiple of the primitive class.","section":"6.2, Theorem 6.3"},{"comment":"The reference [PPZ21] is listed as 'In preparation' while the results used are quoted from [PPZ23]; please update the bibliography.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The defect is localized to the deformation step in Section 6. The local normality argument and the connectedness argument appear coherent and are built on published, parameter-free inputs; I see no circularity. I would be willing to review a revision that replaces the invalid proof of Lemma 6.1 with a correct argument, or supplies a different deformation argument, and that clarifies the algebraic-space and flatness issues in Theorem 6.3."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my honest read. The genuinely new piece is Theorem 1.2/4.10: normality of M_σ(Ku(X), mv0) when v0 is primitive with v0^2 ≥ 5. That goes beyond [FGLZ25b, Prop 5.1], which handles primitive classes, and the proof via Serre-twisted quadratic local models is substantial. I did not find a concrete error in the induction in Prop 4.9, in the rank bound Lemma 4.7, or in the KLS-style connectedness argument in Theorem 5.2. The result is conditional on external inputs (formality [CPZ24], Hodge signature [BP23]) but those are published, parameter-free theorems, so that is not a circularity problem.\n\nThe main problem is Lemma 6.1, the deformation lemma used to get the main theorem. As written, the proof says: take η the generic point of C and Y = M_η. Then 'π|_Y is proper and contains η in its image, so π|_Y dominates C'. That is simply wrong: Y is the fibre over η, so π|_Y has image {η}; it is not a C-scheme and cannot dominate C. Everything after that—Y∩Z nonempty, Y flat, the contradiction—does not follow. Theorem 6.3 uses this lemma to carry irreducibility from the special fibre to a neighborhood, so the main theorem lacks support as written. I think the lemma is probably true: if a component Z were contracted to a, the irreducible special fibre M_a would force Z into the closure of the generic fibre, contradicting that Z is an extra component. But that proof is not the one in the paper, and a referee should require it.\n\nSmaller issue: the abstract says 'we show that the Bridgeland moduli spaces ... are normal', but Theorem 1.2 only proves normality when v0 is primitive with v0² ≥ 5 (or the square-one cases known via identifications). Either the generality should be stated precisely or the theorem should be extended.\n\nNone of this is a takedown. The paper has a real result in it and the main architecture is sound. But because Lemma 6.1 is load-bearing, the paper shouldn't be accepted as is. It deserves a serious referee, and I'd send it out. If the gap is patched and the abstract is corrected, this is a paper I'd cite. For people in derived categories and Fano threefold geometry, this is worth reading even now, but with a warning label on Section 6.","headline":"Genuinely new normality and connectedness results, but the deformation lemma that drives Theorem 6.3 is unproved as written and the abstract oversells the normality claim.","tokens_in":21719,"tokens_out":7446,"would_cite":false,"duration_ms":66239,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14F05","14J45","14D20","14D23"],"pacs":[],"model":"deepseek-v4-flash","headline":"For a general Gushel–Mukai threefold, the Bridgeland moduli space of a primitive odd-square class is irreducible, and all such moduli spaces are normal for primitive classes of square at least five.","keywords":["Derived categories","Bridgeland stability conditions","Kuznetsov components","Gushel–Mukai threefolds","Moduli spaces","Enriques categories","normality","irreducibility"],"falsifier":"A concrete check is to compute the rank of the quadratic Yoneda square $q_S \\colon \\mathrm{Ext}^1_{\\mathcal{C}}(S,S) \\to \\mathrm{Ext}^2_{\\mathcal{C}}(S,S) \\simeq \\mathbb{C}$ at a $\\tau$-fixed stable object $S$ of primitive class $v_0$ with $v_0^2 \\geq 5$; finding rank $1$ or $2$ would break Lemma 4.7 and with it the normality proof. Alternatively, exhibiting a smooth Gushel–Mukai threefold $X$ and a primitive odd-square class $v_0$ for which $M_\\sigma(\\operatorname{Ku}(X),v_0)$ is normal but disconnected would disprove Theorem 5.2 and hence Theorem 6.3.","tokens_in":20704,"feed_emoji":"📐","tokens_out":17062,"duration_ms":120677,"temperature":0.7,"pith_summary":"This paper proves two global facts about the Bridgeland moduli spaces attached to the Kuznetsov component $\\operatorname{Ku}(X)$ of a Gushel–Mukai threefold $X$. First, if $v_0$ is a primitive numerical class with $v_0^2 \\geq 5$, then for every smooth $X$ the moduli space $M_\\sigma(\\operatorname{Ku}(X), m v_0)$ is normal for every positive integer $m$, even though it may contain strictly semistable objects and may be singular. Second, for primitive classes of odd square, a general $X$ has $M_\\sigma(\\operatorname{Ku}(X), v_0)$ irreducible, so the space is a single normal projective variety rather than a union of components. The point is that these spaces are not automatically smooth: stable objects fixed by the Serre involution can be singular points, and the paper shows these singularities are quadratic hypersurfaces of rank at least three, which is precisely enough to preserve normality and to rule out disconnectedness.","feed_headline":"Primitive odd-square moduli on Gushel-Mukai threefolds are irreducible","feed_subtitle":"It proves normality for all primitive classes of square at least five, and one component for odd-square classes.","key_machinery":"The load-bearing object is the Serre-twisted quadratic local model. At a $\\sigma$-polystable object $E \\simeq \\bigoplus_i E_i \\otimes V_i$ of class $m v_0$, the analytic germ of the moduli space is the reductive quotient $Z_d /\\!/ G_d$ of the zero fibre $Z_d = \\mu_d^{-1}(0)$ of a quadratic map $\\mu_d \\colon \\mathrm{Rep}_d \\to \\mathrm{Obs}_d$ built from the Serre pairings $\\mathrm{Ext}^1(E_i,E_j) \\otimes \\mathrm{Ext}^1(E_j,\\tau E_i) \\to \\mathbb{C}$. The dimension formula $a_{ij} = v_0^2 m_i m_j + \\delta_{ij} + \\delta_{j,\\nu(i)}$ for the representation space, together with the rank estimate $\\mathrm{rk}(q_S) \\geq v_0^2/2 \\geq 3$ for the quadratic Yoneda square at a $\\tau$-fixed stable object, turns $Z_d$ into a reduced local complete intersection that is regular in codimension one, hence normal. The second mechanism is a blow-up Chern-class identity: when a normal moduli space of odd-square primitive class is singular and disconnected, elementary modifications of a relative Ext complex on the blow-up of a component force $D^d = 0$ even though $\\deg(D^d) = (-1)^{d-1} \\mathrm{mult}_p(Y) \\neq 0$, a contradiction that proves connectedness.","core_discovery":"The central discovery is a statement of global control: the only singularities that can occur are quadratic hypersurface singularities of rank at least three, and that rank bound is exactly what forces the moduli space to be normal and, in the odd-square primitive case, connected. The main theorem (Theorem 6.3) says that for a fixed primitive numerical class $v_0$ with $v_0^2$ odd, the moduli space $M_\\sigma(\\operatorname{Ku}(X), v_0)$ is irreducible for a general smooth Gushel–Mukai threefold $X$. Its engine is Theorem 4.10: for primitive $v_0$ with $v_0^2 \\geq 5$, the moduli space $M_\\sigma(\\operatorname{Ku}(X), m v_0)$ is normal for every $m \\geq 1$. Since all Serre-invariant stability conditions on $\\operatorname{Ku}(X)$ lie in a single $\\widetilde{\\mathrm{GL}}{}^+_2(\\mathbb{R})$-orbit, the choice of $\\sigma$ is immaterial.","pith_inferences":["The author leaves implicit that the same deformation recipe could address even-square primitive classes: the blow-up argument fails only because the dimension $d = v^2+1$ is odd there, so a parity-sensitive variant might settle connectedness for even-square classes as well.","The deformation lemma is a general recipe: for any family of Bridgeland moduli spaces with normal equidimensional fibres, one irreducible singular fibre forces irreducibility of nearby fibres, so the same strategy could transfer to other Enriques categories.","Because the quadratic singularities have rank at least three, the singular locus has codimension at least two; this suggests these moduli spaces may satisfy stronger local properties, such as local factoriality, or may admit symplectic resolutions in low-dimensional examples.","The role of the Serre-fixed stable locus suggests a general principle for Enriques categories: irreducible components of Bridgeland moduli spaces are controlled by the fixed locus of the involutive autoequivalence on the Calabi–Yau cover."],"forward_implications":["For a general Gushel–Mukai threefold and any primitive odd-square numerical class, the Bridgeland moduli space is a single irreducible normal projective variety of dimension $v_0^2+1$.","Normality holds for all multiples $m v_0$ with $v_0^2 \\geq 5$, including strictly semistable classes, so the moduli space has no embedded components and its singular locus has codimension at least two.","The irreducibility statement is independent of the choice of Serre-invariant stability condition, since all such conditions determine the same stable objects.","The square-one cases, where the moduli spaces are already identified with known moduli spaces, are recovered; the new cases are primitive classes of odd square at least five.","Singular irreducible moduli spaces exist as special fibres in the family, and their irreducibility deforms to the general fibre, producing irreducible moduli spaces with quadratic hypersurface singularities."],"supporting_citations":[{"why":"Establishes the Enriques-category framework: nonemptiness, generic smoothness, the Calabi–Yau cover, and the cover-moduli relation used to construct the singular special fibre.","marker":"[PPZ23]"},{"why":"Supplies formality of derived endomorphism algebras and the quadratic local model identifying the germ of the moduli space with a reductive quotient of a zero fibre.","marker":"[CPZ24]"},{"why":"The blow-up elementary-modification and Chern-class argument whose adaptation proves connectedness of singular normal moduli spaces.","marker":"[KLS06]"},{"why":"Provides relative stability conditions and proper relative moduli spaces in families, used to deform the special irreducible fibre to a general threefold.","marker":"[BLM+21a]"},{"why":"Cited for the dimension, normality, and connectedness background for Bridgeland moduli spaces on Gushel–Mukai varieties, including Proposition 5.1 used in the final irreducibility step.","marker":"[FGLZ25b]"},{"why":"Gives the structure and signature of the Hodge lattice on the Calabi–Yau cover, including Theorem 4.15(3) used in the rank bound.","marker":"[BP23]"},{"why":"Supplies the signature of the real (1,1) Hodge part of the cover, used with [BP23] to conclude that the anti-invariant class has non-positive square.","marker":"[Per19]"},{"why":"Provides the étale slice theorem and good-moduli-space comparison needed to identify simple points of the local zero fibre with stable objects.","marker":"[AHR20]"}],"fun_headline_variants":["Odd-square primitive moduli on Gushel-Mukai threefolds are irreducible","On general Gushel-Mukai threefolds, odd-square primitive moduli are irreducible","Normality for primitive Bridgeland moduli on Gushel-Mukai threefolds","Irreducible Bridgeland moduli for odd-square classes on general GM threefolds","Bridgeland moduli on GM threefolds: normality and odd-square irreducibility"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof leans on the claim that the algebraic Mukai classes on the Calabi–Yau cover that are reversed by the residual involution have non-positive square, because the real $(1,1)$ Hodge intersection form has signature $(2,20)$; if that signature were different, the rank of the quadratic singularities could drop below three and normality would no longer follow.","fun_headline_variants_meta":{"raw":{"variants":["Odd-square primitive moduli on Gushel-Mukai threefolds are irreducible","On general Gushel-Mukai threefolds, odd-square primitive moduli are irreducible","Normality for primitive Bridgeland moduli on Gushel-Mukai threefolds","Irreducible Bridgeland moduli for odd-square classes on general GM threefolds","Bridgeland moduli on GM threefolds: normality and odd-square irreducibility"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001214,"raw_usage":{"total_tokens":4947,"prompt_tokens":843,"completion_tokens":4104,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":459,"completion_tokens_details":{"reasoning_tokens":3989}},"tokens_in":459,"tokens_out":4104,"duration_ms":23872,"temperature":1.0,"reasoning_tokens":3989,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:27:20.656704+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete check is to compute the rank of the quadratic Yoneda square $q_S \\colon \\mathrm{Ext}^1_{\\mathcal{C}}(S,S) \\to \\mathrm{Ext}^2_{\\mathcal{C}}(S,S) \\simeq \\mathbb{C}$ at a $\\tau$-fixed stable object $S$ of primitive class $v_0$ with $v_0^2 \\geq 5$; finding rank $1$ or $2$ would break Lemma 4.7 and with it the normality proof. Alternatively, exhibiting a smooth Gushel–Mukai threefold $X$ and a primitive odd-square class $v_0$ for which $M_\\sigma(\\operatorname{Ku}(X),v_0)$ is normal but disconnected would disprove Theorem 5.2 and hence Theorem 6.3.","supporting_citations":[],"review_version":2}