{"id":"85317c71-c415-4ce4-999f-88d9df437098","arxiv_id":"2502.02143","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A twisted derived category of K3^[n]-type hyper-Kähler varieties is governed by the oriented Markman-Mukai lattice, proven under a primitivity condition and yielding derived equivalences between fine K3 moduli spaces of the same dimension.","lead":"A new conjecture says the twisted derived category of a hyper-Kähler variety of K3^[n]-type is governed by its Markman-Mukai lattice; the paper proves this under numerical constraints. The proof yields a derived equivalence between any two fine moduli spaces of stable sheaves on a projective K3 surface of the same dimension.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.8 Step 2 assumes a primitive class β: the required γ with ⟨η_Y(β),γ⟩=1 need not exist when η_Y(β) is non-primitive, and the hypotheses do not rule this out.","rationale":"The paper has a coherent structure: Conjecture 1.2 is reduced to a lattice-theoretic statement, and the primitive-span condition is used to produce the δ-classes needed for the twisted D-equivalence machinery. The external inputs (Markman's hyperholomorphic bundle, the twisted D-equivalence theorem) are published and are not the main risk. However, the proof of the pivotal Theorem 3.8 contains an unsupported existence assertion. The vector γ with pairing 1 against η_Y(β) is guaranteed only when η_Y(β) is primitive, and the hypotheses do not establish primitivity. Since Theorem 1.4 depends directly on Theorem 3.8, the central result is conditional on repairing this step. The gap may be fixable by choosing γ in the saturation of Zη_Y(β) or by proving that the cyclic-type hypothesis in the footnote follows from the stated assumptions, but as written the proof is incomplete. Hence the verdict should be CONDITIONAL rather than ACCEPT: the paper should be published only after the author either proves η_Y(β) is primitive or supplies a corrected construction that does not require pairing-1 with β.","tokens_in":17153,"tokens_out":35635,"duration_ms":338837,"concrete_test":"In SageMath, take Λ = U^4 ⊕ E8(−1)^2, fix w and δ_w as the standard δ-pair (e.g. w = e1 − (n−1)f1, δ_w = e1 + (n−1)f1), choose a primitive h ∈ w^⊥ ∩ δ_w^⊥ with h^2 = 0 or 2, and set β = 2h, r = 1, m = 1, s = 0 so that ψ(f) = e + 2h is isotropic and primitive. Use the Witt-extension algorithm to seek an integral isometry ψ of Λ ⊕ U satisfying ψ(δ_v) = δ_w for some δ-class δ_v and ψ(f) = e + 2h. If such ψ exists, η_Y(β) is non-primitive and the γ required in Step 2 of Theorem 3.8 does not exist, so the proof gap is real; if no such ψ exists for any h, the concern is vacuous.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The proof of Theorem 3.8, Step 2 (Section 3.5), states: \"Take an integral vector γ ∈ Λ_K3 such that ⟨η_Y(β).γ⟩=1.\" In an even unimodular lattice such a γ exists if and only if η_Y(β) is primitive. The hypotheses of Theorem 3.8 do not imply this. From ψ(f) = re + mβ + sf one only obtains that ψ(f) is primitive, i.e. gcd(r, s, m·div(β)) = 1, and from ψ(δ_v) = δ_w one gets ⟨δ_w, β⟩ = 0; neither condition forces div(η_Y(β)) = 1. For instance, β = 2h with h primitive in w^⊥ ∩ δ_w^⊥ is compatible with isotropy and primitivity of ψ(f). Step 2 uses the pairing-1 condition to define β_k = mη_Y(β) + rkγ and to compute s_k = s + km + rk^2γ^2/2; without it, the construction of the K3 surface S, the coprime shift k, and the fine moduli space M in Step 3 collapse. Theorem 1.4 invokes Theorem 3.8 as its decisive input, so the central claim is not fully proved as written. This is an internal proof gap, separate from the external black boxes flagged by the reader.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes Conjecture 1.2: for hyper-Kähler varieties X and Y of K3^[n]-type, an oriented Hodge isometry between their Markman-Mukai lattices (L(X),v) and (L(Y),w) should induce an equivalence of twisted derived categories D^b(X,[nθ_v]) and D^b(Y,[ϵnθ_w]). It introduces a canonical Brauer class θ_v, a twisted extended Mukai lattice, and proves a lattice-level Hodge isometry (Theorem 1.3). The main new theorem, Theorem 1.4, asserts that Conjecture 1.2 holds when Span(ϕ(v),w) is a primitive sublattice of L(Y); this is deduced from a derived Torelli-type criterion (Theorem 3.8) built on Markman's projectively hyperholomorphic bundles and the twisted D-equivalence theorem of [16]. Sections 4 and 5 apply Theorem 1.4 to moduli spaces of stable objects on K3 surfaces, yielding Theorem 1.5 (twisted derived equivalences between M_w and S^[n]) and Corollary 1.6 (untwisted derived equivalences for fine moduli spaces of the same dimension).","tokens_in":17400,"tokens_out":26894,"duration_ms":252514,"significance":"If the proof were fully correct, the paper would provide substantial evidence for Conjecture 1.2 and would answer a question of Huybrechts on derived equivalences between fine moduli spaces of stable sheaves on K3 surfaces. The formulation of the natural Brauer class θ_v and the twisted extended Mukai lattice is elegant, the lattice-theoretic computations in Section 2 are explicit, and the paper introduces no fitted parameters. These are genuine strengths. However, the central derivation is conditional on substantial external inputs ([12], [16], [17]) and, more seriously, on a step in the proof of Theorem 3.8 that is not justified as written. The claims are therefore not fully established in the present form.","major_comments":[{"comment":"The proof chooses an integral vector γ ∈ Λ_K3 satisfying ⟨η_Y(β),γ⟩ = 1, and then defines v_k using the B-field shift e_{kγ}. The displayed formula for s_k = s + km + rk^2γ^2/2 is valid only when ⟨η_Y(β),γ⟩ = 1, because the definition of e_B in Section 2.3 contains the term ⟨c,B⟩ = km⟨η_Y(β),γ⟩ in the f-coefficient. Existence of such γ is equivalent to primitivity of η_Y(β) in the even unimodular lattice Λ_K3. The hypotheses of Theorem 3.8 imply only that ψ(f) = re + mβ + sf is primitive, i.e. gcd(r, s, m·div(β)) = 1, and that ψ(δ_v) = δ_w gives ⟨δ_w, β⟩ = 0; neither condition forces η_Y(β) to be primitive. The numerical conditions are compatible, for instance, with β = 2h for a primitive class h with h ⊥ δ_w, for which no such γ exists. Without γ, the construction of the K3 surface S, the coprime shift k, and the fine moduli space M in Step 3 collapses. Since Theorem 1.4 invokes Theorem 3.8 directly, this is a load-bearing gap in the proof of the main theorem.","section":"Section 3.5, Theorem 3.8, Step 2"},{"comment":"Theorem 3.1 is stated as a slight generalization of the main result of [16], but its proof is condensed to a reference to Markman's decomposition of parallel transport operators into prime exceptional reflections and birational maps, with no derivation that prime exceptional reflections act by twisted derived equivalences with the stated B-field. Theorem 3.1 is used in Cases 1 and 3 of Theorem 3.8, so this is not a purely cosmetic omission. The proof should be expanded, or the exact statement in [16] that implies the theorem should be quoted.","section":"Section 3.2, Theorem 3.1"}],"minor_comments":[{"comment":"The sentence 'This case is now concluded by Theorem 3.8' is self-referential; it should read 'by Theorem 3.2'.","section":"Section 3.5, end of Step 5"},{"comment":"The notation for inner products is inconsistent: some pairings are written with periods (e.g., ⟨e.f⟩, ⟨δ_v.v⟩) and others with commas (e.g., ⟨η_Y(β),γ⟩). Please unify the notation.","section":"Throughout"},{"comment":"The assertion that x,y ∈ Z follows immediately from primitivity of v, w, and H is not immediate, since the factor 1/k can divide numerators. Please add a sentence explaining why k divides y as well as x.","section":"Section 4.1, Lemma 4.1"},{"comment":"The abstract says 'a natural twisted derived category of any hyper-Kähler variety of K3^[n]-type is controlled by its Markman-Mukai lattice'; this is the conjecture, not an established fact, and the wording should make the conditional status clearer.","section":"Introduction, abstract"}],"recommendation":"major_revision","confidential_remarks":"The gap in Step 2 of Theorem 3.8 is substantive: as written, the proof assumes the existence of a vector γ with pairing 1 against η_Y(β), which is equivalent to primitivity of η_Y(β), and that primitivity is not established by the hypotheses. A revision must either prove that a suitable γ exists in the setting of Theorem 1.4, add and verify an explicit primitivity hypothesis, or supply a different construction of the K3 surface S and the coprime shift k. The paper also relies heavily on [12], [16], and [17], including a paper coauthored by the present author; this is not disqualifying, but the editor may wish to have those inputs checked by a second referee. If the Step 2 gap cannot be repaired within the current framework, the central claim of the paper would not be established."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take. The paper does something real: it formulates a natural conjecture that the twisted derived category of a K3^[n]-type hyper-Kähler variety is controlled by its oriented Markman-Mukai lattice, and proves it under a primitive-span condition. The main theorem, Theorem 1.4, is not a restatement of known results. The canonical Brauer class θ_v and the criterion in Theorem 3.8 are new, and the application to fine moduli spaces (Corollary 1.6) answers a question of Huybrechts in the stated cases. The proof strategy is honest: it builds on Markman's projectively hyperholomorphic bundles and the twisted D-equivalence theorem of [16], and it does not hide that dependency.\n\nThe soft spots are real but not disqualifying. Theorem 3.8 is a long argument with several 'direct computation' steps, and the decisive inputs — [16], Markman's theorems, Yoshioka's numerical condition — are imported rather than reproved. That is acceptable when the inputs are published and independent, but a referee should check the range of applicability, especially condition (12). The paper also has the occasional typo (Step 5 of Theorem 3.8 says 'concluded by Theorem 3.8' where it should refer to Theorem 3.2), and the compressed computations in Section 3.5 deserve scrutiny. The full conjecture remains open; the theorem covers a special but natural case.\n\nOne concern I saw in the stress-test does not hold up. The claim that Step 2 of Theorem 3.8 needs a γ with pairing 1 with η_Y(β) and that this may not exist because η_Y(β) might be non-primitive is not compatible with the text: β is explicitly taken to be a primitive class in H^2(Y,Z), and the marking η_Y is an isometry. Since η_Y(β) lies in Λ_K3, primitivity in the full lattice implies primitivity in Λ_K3, so the unimodularity of Λ_K3 gives the required γ. The worry evaporates on a careful reading.\n\nThe citation pattern is fine. [16] is a coauthored paper, but it is a published independent theorem, and the paper does not derive it from Conjecture 1.2. There is no fitted data or circular reasoning.\n\nThis is a paper for people working on derived categories and moduli of sheaves on hyper-Kähler varieties. It deserves a serious referee. I would send it out and ask for verification of the external inputs and the omitted computations. My own verdict is close to accept, with moderate confidence.","headline":"A serious, genuinely new twisted derived Torelli theorem for K3^[n]-type hyper-Kähler varieties under a primitive-span condition, with one apparent proof gap that dissolves on close reading.","tokens_in":17979,"tokens_out":4875,"would_cite":true,"duration_ms":41391,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14F08","14J28","14J42","14F22"],"pacs":[],"model":"deepseek-v4-flash","headline":"The Markman-Mukai lattice controls a natural twisted derived category of K3^[n]-type hyper-Kähler varieties under a primitivity condition.","keywords":["$K3^{[n]}$-type hyper-Kähler varieties","Markman-Mukai lattice","twisted derived categories","Brauer classes","moduli spaces of stable sheaves on $K3$ surfaces","derived equivalence","Hodge isometry","projectively hyperholomorphic bundles"],"falsifier":"Compute Hochschild homology for a pair satisfying the hypotheses of Theorem 1.4 but with a non-primitive span. Since derived equivalences preserve Hochschild homology, a pair for which these vector spaces differ would show the claimed equivalence fails.","tokens_in":1974,"feed_emoji":"🔗","tokens_out":5850,"duration_ms":123454,"temperature":0.7,"pith_summary":"The paper proposes that every hyper-Kähler variety of $K3^{[n]}$-type carries a canonical Brauer class, and that its twisted derived category is determined by the oriented Markman-Mukai lattice. The main theorem proves this when the span of the relevant lattice vectors is a primitive sublattice. As a consequence, for a projective $K3$ surface and a primitive Mukai vector $w$ with $w^2=2n-2\\ge 2$, the twisted derived category of the moduli space $M_w$ is equivalent to the untwisted derived category of the Hilbert scheme $S^{[n]}$. In the fine case the Brauer classes vanish, answering a question about derived equivalence of fine moduli spaces of stable sheaves on $K3$ surfaces.","feed_headline":"One lattice controls twisted derived categories","feed_subtitle":"A Hodge isometry of the oriented Markman-Mukai lattice forces an equivalence between the natural twisted derived categories.","key_machinery":"The oriented Markman-Mukai lattice $(L(X),v)$ carries the argument: $L(X)$ extends $H^2(X,\\mathbb{Z})$ by a rank-one primitive class $v$ with $v^2=2n-2$, and choosing $v$ fixes an orientation. Associated to $v$ is a canonical class $\\theta_v\\in H^2(X,\\mu_{2n-2})$, defined through auxiliary classes $\\delta_v$ with $\\delta_v^2=2-2n$ and divisibility $2n-2$; the Brauer class of $\\theta_v$ is the obstruction to $X$ being a fine moduli space. The paper also introduces the twisted extended Mukai lattice $\\widetilde{H}(X,\\delta_v/(2n-2),\\mathbb{Z})\\cong H^2(X,\\mathbb{Z})\\oplus U$, and uses integral isometries called Eichler transvections to move the orientation vector into a convenient position. The decisive criterion, Theorem 3.8, states that an orientation-preserving Hodge isometry between twisted extended Mukai lattices that sends a $\\delta$-class to a $\\delta$-class is realized by a twisted derived equivalence; this is powered by projectively hyperholomorphic bundles and by the twisted D-equivalence theorem for birational $K3^{[n]}$-type varieties.","core_discovery":"The central claim is Conjecture 1.2: if $X$ and $Y$ are hyper-Kähler varieties of $K3^{[n]}$-type and there is a Hodge isometry $\\phi:(L(X),v)\\to (L(Y),w)$ between oriented Markman-Mukai lattices, then $D^b(X,[n\\theta_v])\\cong D^b(Y,[\\epsilon n\\theta_w])$, where $\\epsilon=\\pm1$ according as $\\phi$ preserves or reverses orientation. Here $L(X)$ is the Markman-Mukai lattice, an extension of $H^2(X,\\mathbb{Z})$ by a primitive generator $v$ of square $2n-2$, and $\\theta_v$ is the canonical class in $H^2(X,\\mu_{2n-2})$ determined by that orientation. The paper proves the conjecture when $\\operatorname{Span}(\\phi(v),w)\\subset L(Y)$ is a primitive lattice embedding. The proof builds a Hodge isometry of twisted extended Mukai lattices and then uses a projectively hyperholomorphic bundle construction together with the twisted D-equivalence theorem to upgrade this isometry to a twisted derived equivalence. The main application is that for a projective $K3$ surface $S$ and a primitive class $w$ with $w^2=2n-2\\ge 2$, one has $D^b(M_w,[n\\theta_w])\\cong D^b(M_w,[-n\\theta_w])\\cong D^b(S^{[n]})$.","pith_inferences":["If Conjecture 1.2 holds without the primitivity hypothesis, then the full twisted derived Torelli theorem for $K3^{[n]}$-type would follow, suggesting that the conjectural $K3$ category associated to a hyper-Kähler variety is itself determined by the oriented Markman-Mukai lattice.","The same lattice-governed mechanism may extend to moduli spaces of Bridgeland-stable objects on arbitrary $K3$ categories, with the canonical Brauer class controlling which twists appear when the moduli space is not fine.","The boundary of the theorem could be mapped by testing explicit pairs with non-primitive spans, for instance low-dimensional examples where the twisted derived equivalence can be checked by direct Fourier–Mukai kernels."],"forward_implications":["Theorem 1.4 gives a twisted derived Torelli statement under the primitivity hypothesis: a Hodge isometry of oriented Markman-Mukai lattices forces an equivalence of the corresponding twisted derived categories.","For any projective $K3$ surface and primitive Mukai vector $w$ with $w^2=2n-2\\ge 2$, the twisted derived category of $M_w$ is derived equivalent to the Hilbert scheme $S^{[n]}$.","When $\\operatorname{div}(w)=1$, the Brauer class vanishes and $D^b(M_w)\\cong D^b(M_v)$ for any fine $v$ with the same square; in particular, any two fine moduli spaces of stable sheaves on a $K3$ surface of the same dimension are derived equivalent.","The criterion in Theorem 3.8 provides a reusable template for lifting Hodge isometries of twisted extended Mukai lattices to twisted derived equivalences for other $K3^{[n]}$-type pairs."],"supporting_citations":[{"why":"Supplies the twisted D-equivalence theorem, the main input that turns Hodge parallel transports into twisted derived equivalences.","marker":"[16]"},{"why":"Provides the projectively hyperholomorphic bundle and the universal bundle with its rational Hodge isometry.","marker":"[12]"},{"why":"Supplies the structural properties of the Markman-Mukai lattice and the parallel-transport formalism.","marker":"[10]"},{"why":"Gives the numerical condition ensuring the moduli space consists of stable vector bundles.","marker":"[17]"},{"why":"Defines the analogue of the canonical theta class and the fine-moduli-space criterion.","marker":"[11]"},{"why":"Provides the obstruction-theoretic criterion for fine moduli spaces and the untwisted counterexample.","marker":"[15]"},{"why":"Witt's theorem is used to extend isometries of primitive sublattices to the full lattice.","marker":"[8]"}],"fun_headline_variants":["One lattice controls twisted derived categories","Twisted derived equivalence pinned by a single lattice","Markman-Mukai lattice dictates twisted D-branes","Lattice twist: derived categories unified by isometry","Proof: lattice fixes twisted derived equivalence"],"cache_read_input_tokens":20096,"weakest_assumption_plain":"The proof relies, without reproving, on the twisted D-equivalence theorem and Markman's projectively hyperholomorphic bundle theorem; if either fails in the range used here, the derived equivalences in Theorem 1.4 and Corollary 1.6 do not follow.","fun_headline_variants_meta":{"raw":{"variants":["One lattice controls twisted derived categories","Twisted derived equivalence pinned by a single lattice","Markman-Mukai lattice dictates twisted D-branes","Lattice twist: derived categories unified by isometry","Proof: lattice fixes twisted derived equivalence"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000253,"raw_usage":{"total_tokens":1566,"prompt_tokens":952,"completion_tokens":614,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":568,"completion_tokens_details":{"reasoning_tokens":545}},"tokens_in":568,"tokens_out":614,"duration_ms":6556,"temperature":1.0,"reasoning_tokens":545,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T13:15:04.233432+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute Hochschild homology for a pair satisfying the hypotheses of Theorem 1.4 but with a non-primitive span. Since derived equivalences preserve Hochschild homology, a pair for which these vector spaces differ would show the claimed equivalence fails.","supporting_citations":[{"cited_title":"The D-equivalence conjecture for hyper- k¨ ahler varieties via hyperholomorphic bundles.Invent","cited_arxiv_id":null,"evidence_quote":"Supplies the twisted D-equivalence theorem, the main input that turns Hodge parallel transports into twisted derived equivalences."},{"cited_title":"Rational Hodge isometries of hyper-K¨ ahler varieties ofK3[n] type are algebraic","cited_arxiv_id":null,"evidence_quote":"Provides the projectively hyperholomorphic bundle and the universal bundle with its rational Hodge isometry."},{"cited_title":"A survey of Torelli and monodromy results for holomorphic-symplectic varieties","cited_arxiv_id":null,"evidence_quote":"Supplies the structural properties of the Markman-Mukai lattice and the parallel-transport formalism."},{"cited_title":"Stability and the Fourier-Mukai transform","cited_arxiv_id":null,"evidence_quote":"Gives the numerical condition ensuring the moduli space consists of stable vector bundles."},{"cited_title":"The Beauville-Bogomolov class as a characteristic class","cited_arxiv_id":null,"evidence_quote":"Defines the analogue of the canonical theta class and the fine-moduli-space criterion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Witt's theorem is used to extend isometries of primitive sublattices to the full lattice."}],"review_version":1}