{"id":"5db714e9-24f4-4988-b781-9962f15d3a1a","arxiv_id":"2412.12903","paper_version":3,"verdict":"CONDITIONAL","confidence":"LOW","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Quiver Hecke algebra module categories of arbitrary symmetrizable type admit Kleshchev stratifications, and quantum-unipotent subcategories are affine highest weight categories.","lead":"This paper proves that module categories over quiver Hecke algebras of arbitrary type can be organized into many highest-weight-like layerings, and that the subcategories attached to quantum unipotent subgroups are affine highest weight categories. The result extends earlier proofs for finite and symmetric affine types to all symmetrizable types over any field, using explicit standard modules built from determinantial modules and R-matrix methods.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Definition 5.14 applies the finite-dimensional duality D to infinite-dimensional standard modules, so the proper costandard modules are not defined as finite-dimensional modules.","rationale":"The paper's central claim—that R(β)-gMod is stratified—depends on the criterion Proposition 5.7, which requires a genuine proper costandard module ∇(λ)∈H-gmod for each simple. Definition 5.14 attempts to produce these as D(∆(λ,s)) using the finite-dimensional duality D, but the inputs ∆(λ,s) are infinite-dimensional. This is a concrete, textually verifiable gap: D is only defined on R-gmod in Section 3.1, and no extension to R-gMod is supplied. The naive extension fails to produce finitely generated modules. This concern is load-bearing because Theorem 5.17, Proposition 5.16, and hence Theorem 5.18 all rely on the existence of finite-dimensional proper costandard modules. The issue is repairable: one could introduce a duality on the appropriate category of finitely generated modules (e.g., a graded Matlis dual adapted to Laurentian modules) or redefine ∇(λ,s) differently, as the reader suggests. The reader's formal weakest_assumption was the injectivity of renormalized R-matrices and the spectral condition in Lemma 3.12; that point also deserves scrutiny, but I find the duality gap more immediately visible and more directly tied to the main theorem's proof. The reader did note the duality problem in the rationale, so agreement is partial. My verdict remains CONDITIONAL, matching the reader's verdict; I do not see a reason to change it, though the duality gap strengthens the case for requesting a revision before full acceptance.","tokens_in":80554,"tokens_out":6892,"duration_ms":64252,"concrete_test":"Work in type A1 with w=s1 and β=α1. Then ∆(1)=ˆL(β1)^(1)=R(α1)≅k[x], which is infinite-dimensional. The formula ∇(1)=D(∆(1)) would give Hom_k(k[x],k), an infinite-dimensional vector space with the opposite grading, which is not a finitely generated R(α1)-module and does not lie in R(α1)-gmod. Check whether the paper defines any duality on R-gMod that makes D(∆(1)) a finite-dimensional module; if not, Definition 5.14 fails on this example and Theorem 5.17 requires adjustment.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 3.1, the duality D(M)=Hom_k(M,k) is introduced only on R-gmod, the category of finite-dimensional graded modules. Definition 5.14 defines the proper costandard module by ∇(λ,s)=D(∆(λ,s)). But ∆(λ,s)=∆(s)∘∆(λ) and ∆(λ)=ˆL(β_l)^(λ_l)∘⋯∘ˆL(β_1)^(λ_1), where each affinization ˆL(β_k) is infinite-dimensional over k (e.g., free of finite rank over k[z] by Lemma 3.14). Hence D is not defined on ∆(λ,s). If one naively interprets D as the restricted dual ⊕_d Hom_k(M_{-d},k), then for a bounded-below infinite-dimensional module the result is bounded above and is not finitely generated, so it does not belong to R-gmod and is not finite-dimensional. This contradicts Definition 5.4, which explicitly requires proper costandard modules to be finite-dimensional. Theorem 5.17 asserts that these ∇(λ,s) are the proper costandard modules, and Proposition 5.16 and the criterion Proposition 5.7 rely on their existence and finiteness. Without a valid definition of ∇(λ,s) satisfying Definition 5.4, the proof of the stratification Theorem 5.18 is not complete. This is a load-bearing definitional gap, not a notational slip: the Ext orthogonality and BGG-reciprocity computations in Lemma 5.15 and Proposition 5.16 depend essentially on ∇(λ,s) being a finite-dimensional module with the asserted properties.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a new algebraic approach to proving that, for arbitrary symmetrizable root data and arbitrary base fields, the category R(β)-gMod of finitely generated graded modules over a quiver Hecke algebra admits a Kleshchev stratification, and that the full subcategory R_{w,*} associated with a quantum unipotent subgroup is affine highest weight. The standard modules are realized as convolutions of affinized determinantial modules, and the proofs rely on R-matrix techniques, including a foundational short exact sequence for affinizations and an Ext-vanishing theorem. The paper also treats affine Lie types and identifies certain standard modules with imaginary root vectors.","tokens_in":80821,"tokens_out":4311,"duration_ms":42606,"significance":"If the main theorems are established, this is a substantial advance: it extends prior results of Kato, Brundan-Kleshchev-McNamara, McNamara, and Kleshchev-Muth from finite or symmetric affine types to arbitrary symmetrizable types and arbitrary characteristic, with explicit algebraic descriptions of standard and proper costandard modules. The paper is well-structured, gives many detailed proofs, and introduces a promising use of R-matrices for homological questions. However, the current manuscript contains a load-bearing definitional gap involving the duality functor, which prevents the main theorem from being proven as written. The central ideas appear plausible and the gap is likely fixable, but a revision is necessary.","major_comments":[{"comment":"The duality functor D is introduced in Section 3.1 only on the category R-gmod of finite-dimensional graded modules. Definition 5.14 sets ∇(λ,s) = D(∆(λ,s)), where ∆(λ,s) = ∆(s)∘∆(λ) is built from affinizations ˆL(β_k) that are free of finite rank over k[z] (Lemma 3.14) and hence are infinite-dimensional over k. Similarly, Definition 5.12 sets ∇(λ) = D(∆(λ)) for the truncated standard modules. Thus D is not defined on these objects. If one interprets D as the restricted dual, ∇(λ,s) is not finite-dimensional and does not belong to R-gmod, contradicting Definition 5.4, which explicitly requires proper costandard modules to be finite-dimensional. This invalidates the claimed existence of proper costandard modules in Theorem 5.17 and undermines the proofs of Lemma 5.15, Proposition 5.16, and the main stratification Theorem 5.18, since the criterion Proposition 5.7 requires finite-dimensional proper costandard modules.","section":"§5.4, Definition 5.14 (and §5.3, Definition 5.12)"},{"comment":"The proof of the foundational short exact sequence relies on the exactness of the functor F constructed from a type-A₂ quiver Hecke algebra. The paper asserts that F is exact because the global dimension of R_{A₂}(α)-gMod is finite, citing [KKK18, Proposition 3.7] and [KKOP24, Proposition 7.6], but then states that for the needed cases α ∈ {0, α₁, α₂, α₃} one can directly verify finiteness without providing that verification. Since the exactness of F is essential for producing the short exact sequence for arbitrary w and i, this step needs a complete argument rather than a promise.","section":"§4.3, Proof of Theorem 4.11"}],"minor_comments":[{"comment":"The word 'Definintion' appears in place of 'Definition' in several places (e.g., Definitions 2.2, 2.3, 3.1, 3.8, 3.15, 3.23, 3.29, 3.35, 3.41, 3.54, 4.21, 5.4, 6.17).","section":"Throughout"},{"comment":"The notation ∆(λ,s) is used with two different meanings: in Section 3.7 it denotes L(s)∘L(λ_lβ_l)∘⋯∘L(λ₁β₁), while in Definition 5.14 it denotes ∆(s)∘∆(λ). Although the latter is claimed to coincide with the standard module, the reuse of the same symbol for a priori different objects is confusing and should be clarified.","section":"§3.7 and §5.4"},{"comment":"The proof of Lemma 2.5(3) uses the notation M^{n'} and then concludes ext^k_H(M^{n'},N)=0, but the transition from the construction of the projective resolution of M^{n'} to the vanishing is terse; a few more details would improve readability.","section":"§2.2, Lemma 2.5"}],"recommendation":"major_revision","confidential_remarks":"The definitional gap concerning the duality functor is central and currently prevents the main theorem from being valid as written. That said, the gap appears fixable by redefining proper costandard modules in a way that avoids applying D to infinite-dimensional modules, perhaps through a finite-dimensional truncation or a dual construction in a different category, and then re-proving the required Ext orthogonality and BGG-style reciprocity. The paper is ambitious and likely correct after such a revision, but as it stands the main claim is not established."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi, here is my read on 2412.12903. The headline: this is the right paper at the right time. The theorem—stratification of R(β)-gMod for arbitrary symmetrizable type, over any field—would close a long gap. The R-matrix method is genuinely different, and the explicit identification of standard modules via affinized determinantial modules is new and likely useful. The paper is careful with the literature; it even corrects an error in the proof of a determinantial-module fact from KKOP18. No circularity: the main theorem is not assumed, and the dependence on external results is heavy but legitimate.\n\nThat said, the stress test is right about Definition 5.14. Section 3.1 defines D only on R(β)-gmod, the finite-dimensional graded modules. Definition 5.14 sets ∇(λ,s) = D(∆(λ,s)), but ∆(λ,s) is built from affinizations, which are infinite-dimensional over k (free of finite rank over k[z]). So D, as defined, does not apply. The restricted dual would be bounded above and not finitely generated, so it would not land in R-gmod and would not satisfy Definition 5.4’s requirement that proper costandard modules be finite-dimensional. This is not a typo: Lemma 5.15 and Proposition 5.16 use Ext orthogonality and BGG reciprocity for these objects, and Proposition 5.7’s criterion needs genuine finite-dimensional proper costandard modules. Without a valid definition of ∇(λ,s), the proof of Theorem 5.18 is incomplete.\n\nIs it fatal? I don’t think it has to be. It looks repairable: define the proper costandard modules directly, or extend the duality to a suitable subcategory of infinite-dimensional modules and prove finiteness afterwards. The author should also fix the cross-reference in the proof of Theorem 3.58(2)—it cites Proposition 5.5, which seems to be the wrong result.\n\nThe reader’s other worry—about Proposition 3.20 depending on Lemma 3.12’s spectral condition—is less serious. Lemma 3.12 proves that condition from the affinization axioms; it is not an unstated assumption. So the R-matrix core seems solid.\n\nVerdict: I would send this to a serious referee. The result is important enough, and the framework is sound enough overall. The referee should be asked to check whether the duality gap can be closed without changing the main statements.","headline":"A genuinely important theorem with a clean R-matrix approach, but the proper costandard modules are not defined as written because the duality D is only set up for finite-dimensional modules.","tokens_in":81403,"tokens_out":3912,"would_cite":false,"duration_ms":38137,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["17B37","17B67","16G99"],"pacs":[],"model":"deepseek-v4-flash","headline":"Module categories of quiver Hecke algebras are stratified in arbitrary type and over any field, with standard modules built from determinantial modules.","keywords":["quiver Hecke algebra","affine highest weight category","stratified category","determinantial module","R-matrix","categorification","quantum unipotent subgroup","graded modules"],"falsifier":"Take the affinization $\\hat M(w\\Lambda_i,\\Lambda_i)$ constructed after Proposition 3.51 and compute the action of $p_{i,\\beta}$ on it: Lemma 3.12 requires this action to be a nonzero power of $z$. If there exists some $w$ and field where $p_{i,\\beta}$ acts by zero, then the injectivity of renormalized R-matrices (Proposition 3.20) fails at that step, and the short exact sequence of Theorem 4.11 would need another proof. Since Proposition 3.20 is used to prove that $\\hat M_2$ is a projective cover (Theorem 4.16), checking this spectral condition in unexplored types or characteristics would directly settle whether the stratification extends.","tokens_in":80286,"feed_emoji":"🧶","tokens_out":6396,"duration_ms":59178,"temperature":0.7,"pith_summary":"The paper proves that, for every symmetrizable root datum and every base field, the category of finitely generated graded modules over a quiver Hecke algebra admits a stratification: the simple modules can be ordered so that each projective cover has a filtration by explicit standard modules. The standard modules are realized as convolution products of affinizations of determinantial modules, which the paper identifies as the categorical counterpart of the PBW basis. A truncation of this result shows that the full subcategory attached to any Weyl group element—the categorical analogue of a quantum unipotent subgroup—is an affine highest weight category with polynomial endomorphism rings and finite global dimension. If correct, this unifies and extends earlier special cases covering finite and symmetric affine types, and it does so without case-by-case computation. The argument is carried by the injectivity of renormalized R-matrices and by a short exact sequence that relates three determinantial modules.","feed_headline":"Quiver Hecke module categories stratify in every type and field","feed_subtitle":"Standard modules built from affinized determinantials yield affine highest weight categories.","key_machinery":"The central objects are determinantial modules $M(w\\Lambda,v\\Lambda)$ and their affinizations $\\hat M(w\\Lambda,v\\Lambda)$: graded modules over the quiver Hecke algebra that categorify unipotent quantum minors. The mechanism that carries the argument is the renormalized R-matrix: Proposition 3.20 asserts that renormalized R-matrices of affinized determinantial modules are injective, and Theorem 4.11 uses this injectivity to construct the short exact sequence relating $\\hat M(w\\Lambda_i,\\Lambda_i)$, $\\hat M(w s_i\\Lambda_i,w\\Lambda_i)$, and $\\hat M(w s_i\\Lambda_i,\\Lambda_i)$. This converts the braider structure of the convolution product into homological information: Theorem 4.16 shows that $\\hat M(w s_i\\Lambda_i,w\\Lambda_i)$ is the projective cover of its head, which together with the BGG-type formula gives the stratification.","core_discovery":"The central claim is Theorem 5.18: with respect to the map $\\rho\\colon\\Sigma(\\beta)\\to P_{\\preceq}(\\beta)$ and the partial order $\\le$, the category $R(\\beta)$-gMod is a stratified category. Its standard modules are $\\Delta(\\lambda,s)=\\Delta(s)\\circ\\Delta(\\lambda)$, where $\\Delta(\\lambda)$ is a convolution of affinized determinantial modules and $\\Delta(s)$ is a projective cover in the $R_{\\ast,w}$ subcategory; the proper costandard modules are the duals $\\nabla(\\lambda,s)=D(\\Delta(\\lambda,s))$. Restricting to the subcategory $R_{w,\\ast}(\\beta)$-gMod yields Theorem 5.21: it is an affine highest weight category with standard modules $\\Delta(\\lambda)$ and endomorphism rings isomorphic to polynomial algebras. The proof identifies standard modules with convolutions of affinized determinantial modules, shows $\\operatorname{Ext}^1$-vanishing through a foundational short exact sequence of the form $0\\to q^{(\\alpha_i,\\alpha_i)+(\\gamma_1,\\gamma_2)}\\hat M_1\\circ\\hat M_2\\to\\hat M_2\\circ\\hat M_1\\to\\hat M_3\\to 0$, and verifies the BGG-type reciprocity formula for projective modules using the coincidence of the Ext bilinear form with Kashiwara's bilinear form under categorification.","pith_inferences":["The same R-matrix injectivity criterion could be tested on other monoidal categorifications, for example higher-level or exotic analogues, to see whether explicit determinantial standard modules exist there as well.","Because the standard modules are now explicitly described, one can attempt concrete computations of $\\operatorname{Ext}$-algebras between standard modules in types and characteristics where nothing was known before, which would give the first homological data for non-semisimple quiver Hecke categories outside finite and symmetric affine types.","The paper leaves open which simple modules admit an affinization satisfying the spectral condition on the central elements $p_{i,\\beta}$; a counterexample in an unexplored type would isolate exactly where the stratification machinery would need to be modified.","In affine type $A_{2l}^{(2)}$, the exceptional case handled separately in Section 6 offers a concrete test: verifying the two short exact sequences of Theorem 6.22 in low-rank examples would confirm the pattern predicted by the general stratification.","The explicit nature of the standard modules may allow the stratification to be sheared into finer stratifications indexed by other convex orders, connecting the resulting homological invariants to cluster structures on quantum coordinate rings."],"forward_implications":["For arbitrary symmetrizable type and any base field, every finitely generated graded module category $R(\\beta)$-gMod is stratified, so every module has a well-behaved filtration by the explicit standard modules and BGG-style reciprocity holds.","For each Weyl group element $w$, the subcategory $R_{w,\\ast}(\\beta)$-gMod is an affine highest weight category, which implies it has finite global dimension and its standard modules have polynomial endomorphism rings.","The Grothendieck group of $R_{w,\\ast}$-gproj becomes an algebra isomorphic to the quantum unipotent subgroup $U_q(n_-\\cap wn_+)$, giving explicit modules that categorify the PBW and dual PBW bases.","In affine types the stratification can be refined for any convex order, and the standard module corresponding to the minimal imaginary root vector categorifies the imaginary root vector of the affine PBW basis.","The $\\operatorname{Ext}^k$-vanishing for affinized determinantial modules obtained in the paper holds for all $k\\ge 1$, strengthening earlier extension-vanishing results that were only known in special types."],"supporting_citations":[{"why":"Supplies determinantial modules, their affinizations, cuspidal decompositions, and the quantum unipotent subcategories $R_{w,\\ast}$, $R_{\\ast,w}$ used throughout the proof.","marker":"[KKOP18]"},{"why":"Provides the foundational theory of affinizations and renormalized R-matrices for quiver Hecke algebras, including the key injectivity statements the paper refines.","marker":"[KP18]"},{"why":"Gives the monoidal categorification framework, unipotent quantum minors, and the earlier R-matrix results that the paper extends to arbitrary type and field.","marker":"[KKKO18]"},{"why":"Supplies the generalized affinization concept and the duality data used in Step 2 of the proof of Theorem 4.11 to build the exact functor $F$.","marker":"[KKOP24]"},{"why":"Establishes the affine highest weight structure for finite-type quiver Hecke algebras, the case the present paper generalizes.","marker":"[BKM14]"},{"why":"Provides the definition of affine highest weight categories and stratified categories, as well as the criterion used to prove the stratification.","marker":"[Kle15a]"},{"why":"Gives the stratification in symmetric affine type, whose techniques are compared and replaced by R-matrix methods in the present paper.","marker":"[McN17]"},{"why":"Supplies the 2-representation theory of the 2-categorical $U_q(\\mathfrak{sl}_2)$ used to prove Proposition 1.5 on $\\operatorname{Ext}^1$-vanishing.","marker":"[Rou08]"}],"fun_headline_variants":["Affine highest weight categories from quiver Hecke algebras","Every quiver Hecke type yields affine highest weight categories","Stratifying all quiver Hecke module categories","New highest weight categories via determinantial modules","Quiver Hecke modules: affine highest weight structure in all types"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof rests on the injectivity of renormalized R-matrices for affinized determinantial modules, which in turn assumes the central elements $p_{i,\\beta}$ act on the affinizations as nonzero powers of the degree-shift endomorphism $z$; if that spectral condition fails for any determinantial affinization used in Section 3.7, the projectivity of the standard modules and hence the stratification would not be established.","fun_headline_variants_meta":{"raw":{"variants":["Affine highest weight categories from quiver Hecke algebras","Every quiver Hecke type yields affine highest weight categories","Stratifying all quiver Hecke module categories","New highest weight categories via determinantial modules","Quiver Hecke modules: affine highest weight structure in all types"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001525,"raw_usage":{"total_tokens":6093,"prompt_tokens":920,"completion_tokens":5173,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":536,"completion_tokens_details":{"reasoning_tokens":5094}},"tokens_in":536,"tokens_out":5173,"duration_ms":35444,"temperature":1.0,"reasoning_tokens":5094,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:39:11.227461+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the affinization $\\hat M(w\\Lambda_i,\\Lambda_i)$ constructed after Proposition 3.51 and compute the action of $p_{i,\\beta}$ on it: Lemma 3.12 requires this action to be a nonzero power of $z$. If there exists some $w$ and field where $p_{i,\\beta}$ acts by zero, then the injectivity of renormalized R-matrices (Proposition 3.20) fails at that step, and the short exact sequence of Theorem 4.11 would need another proof. Since Proposition 3.20 is used to prove that $\\hat M_2$ is a projective cover (Theorem 4.16), checking this spectral condition in unexplored types or characteristics would directly settle whether the stratification extends.","supporting_citations":[],"review_version":1}