{"id":"2a0635d2-4c0d-453d-850d-964adbee0470","arxiv_id":"2504.14844","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Irreducible components of multiparameter persistence module varieties carry a Kashiwara crystal structure, with explicit operators in the 1- and 2-parameter cases.","lead":"This paper shows that the irreducible building blocks of multiparameter persistence modules can be arranged into a Kashiwara crystal, a combinatorial structure borrowed from quantum group theory. It gives explicit formulas for the crystal operators in the one- and two-parameter cases, and proves the resulting crystal is genuinely different from the standard one.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 5.5 gives invalid Kashiwara operators at boundary weights: for ν=(0,1,1,0), Λ=(ν;0,0), ε₁(Λ)=0 so ~e₁ should be 0, but the formula outputs (ν+α₁;−1,0), not an element of BC.","rationale":"The reader's weakest_assumption was the smoothness/connected-fiber assertion in Proposition 3.8. I examined Proposition 3.8 in detail and found the arguments convincing: the fiber description of ϖ₁ over S(νᵢ−1,1)×∏GL is a vector bundle with fiber gExtᵢ(¯f), the local trivialization using a fixed complement X is algebraic, and Lemma 3.7 gives smoothness; similarly for ϖ₂, the principal GL-bundle description over ∏GL×((k^ε)*\\{0}) is correct. Thus Theorem 3.13 likely holds. Instead, the most concrete and checkable flaw is Theorem 5.5, which fails at boundary weights: the case ν₁=0 gives an output with a negative rank parameter and a weight outside −Q₊. This is not a mere typo: it is a false statement of the explicit operators. Since the abstract and introduction advertise explicit descriptions as a contribution, and Corollary 5.10 and the connectivity claim derive from Theorem 5.5, the paper as written overstates its results. The issue is fixable by adding nonnegativity conditions or explicit zero cases, so the correct verdict remains CONDITIONAL: accept only after Theorem 5.5 (and its dependent corollaries) are corrected and the boundary behavior is described. The reader's verdict of CONDITIONAL is therefore retained, but for a different reason than the one identified.","tokens_in":27924,"tokens_out":32971,"duration_ms":263647,"concrete_test":"Compute ~e₁ for ν=(0,1,1,0), Λ=(ν;0,0) directly from Definition 3.11 and Proposition 3.8: verify that ε₁(Λ)=0 so the operator must be 0, and note that Theorem 5.5(1) would require r₁−1=−1, an invalid rank. Then re-run the same check on the words in Corollary 5.10(1), tracking all intermediate dimension vectors and rank parameters with the condition that each operator is 0 if the output weight is outside −Q₊ or if (r₁,r₂) fails the bounds of Corollary 5.4.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central construction (Theorem 3.13) appears sound: Proposition 3.8 is carefully argued, and the smoothness/connected-fiber hypotheses hold with plausible local trivializations using gExt_i and the fixed-complement argument. The crystal axioms then follow by the standard correspondence argument. The load-bearing problem is in the explicit description, Theorem 5.5, which is false as stated. Take the 2×2 grid with dimension vector ν=(0,1,1,0), so ν₁+ν₄=0≤ν₂+ν₃=2. The unique component is Λ=(ν;0,0). Since vertex 1 has no incoming arrows, ε₁(Λ)=ν₁=0, so Definition 3.11 gives ~e₁Λ=0 (and ϖ₂⁻¹(Λ₂)=∅ by Proposition 3.8(3)). However, Theorem 5.5(1) in the case ν₁+ν₄≤ν₂+ν₃ outputs (ν+α₁; r₁−1, r₂) = (ν+α₁; −1,0). The weight ν+α₁ = α₁−α₂−α₃ is not in −Q₊ (coefficient of α₁ is positive), and the rank r₁−1 is negative, so the output is not an admissible irreducible component. This is a concrete counterexample to Theorem 5.5. The flaw propagates to the claimed proofs of Corollary 5.10(1)–(3) and the connectivity statement, which invoke Theorem 5.5 on components such as u_C where boundary cases can occur. The theorem needs extra clauses: each operator must return 0 whenever the resulting dimension vector has a negative entry or the new rank parameters fall outside their admissible ranges. As written, the paper's advertised 'explicit descriptions' are incorrect, and the nontriviality consequences are not rigorously established, even though the main crystal-existence theorem may survive.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a Kashiwara crystal structure on BC, the set of all irreducible components of the representation varieties of d-parameter persistence modules on the equioriented commutative grid quiver. The main theorem (Theorem 3.13) is proved by direct verification of the crystal axioms, following the geometric strategy of Kashiwara and Saito but with commutative relations instead of preprojective relations. The paper then gives explicit descriptions of the Kashiwara and star operators in the one-parameter and 2x2 cases, parametrizes the irreducible components for the 2x2 grid, and derives structural consequences including non-embeddability into B(∞), failure of upper seminormality, and connectedness. An appendix describes the general module in each irreducible component of the 2x2 case.","tokens_in":28245,"tokens_out":21573,"duration_ms":169191,"significance":"If the main theorem is correct, the paper gives a new geometric/combinatorial invariant for multiparameter persistence modules and a genuine extension of the Kashiwara-Saito construction to a commutative-relation setting. The proof of Proposition 3.8 is detailed and the smoothness and connected-fiber arguments are plausible; I did not find an error in the central geometric construction. The explicit formulas of Theorems 5.5 and 5.6 are a major advertised contribution, and those formulas are currently incorrect at boundary weights. Because the nontriviality corollaries are deduced from those formulas, the paper is not ready for publication in its present form.","major_comments":[{"comment":"The explicit formula for ~e1 is false at boundary weights. Take ν=(0,1,1,0) and Λ=(ν;0,0), the unique irreducible component of EC(ν) by Corollary 5.4(2). Here ν1+ν4=0≤ν2+ν3=2, so the first clause of Theorem 5.5(1) returns (ν+α1;−1,0). This is not an element of BC: ε1(Λ)=ν1=0, so Definition 3.11(2) and Proposition 3.8(3) force ~e1Λ=0; moreover the output has a negative rank parameter r1−1, and if ν+α1 is read as a weight it is not in −Q+. The same boundary defect appears in Theorem 5.6(1) for ~e*_4: for the same Λ, the first clause returns (ν+α4;0,−1), whereas ε*_4(Λ)=ν4=0 forces ~e*_4Λ=0. The stated formulas therefore need explicit zero clauses whenever the resulting dimension vector has a negative entry or the new rank parameters fall outside their admissible ranges, and the proofs of the affected clauses must be reworked.","section":"Section 5, Theorem 5.5(1) and Theorem 5.6(1)"},{"comment":"The numerical values in the proof contradict Proposition 5.8(2). For ν=−α1−α2−α3−2α4, i.e. dimension vector (1,1,1,2), and Λ=(ν;1,1), Proposition 5.8(2) gives ε1(Λ)=1 and, since ν4=2>r2=1, ε′1(Λ)=ν1−r1=0. The proof states that ε1(Λ)=1 and ε′1(Λ)=1, which is inconsistent with the proposition just proved. The conclusion that BC is not upper seminormal may still be salvageable using the corrected value 0, but the printed argument is not valid as it stands.","section":"Section 5, proof of Corollary 5.10(2)"},{"comment":"Both proofs invoke Theorem 5.5 without checking that the operators are applied only inside their valid domains. In particular, the connectedness proof displays a composition of ~e operators that can pass through components with zero dimension at an active vertex, exactly the boundary situation where the current formulas fail. Since Theorem 5.5 is false as stated, the equality used in the non-embedding argument and the connectivity claim are not established. These corollaries must be re-derived after the operator formulas are corrected.","section":"Section 5, Corollary 5.10(1) and (3)"}],"minor_comments":[{"comment":"Proposition 3.8(3) states that the image of ϖ2 is the 'closed subset' of EC(ν) consisting of points f with εi(f)>0. The condition εi(f)>0 is open in the Zariski topology, so this should presumably read 'open subset' or simply 'subset'. The surrounding arguments do not seem to use closedness, but the statement as written is false in general.","section":"Section 3, Proposition 3.8(3)"},{"comment":"In the displayed formula for ~f*_i, the condition 'if νi≥νi' is tautological and is presumably a typo for 'if νi≥νi+1', matching the condition in the preceding line for ~e*_i.","section":"Section 4, Theorem 4.1(2)"},{"comment":"The formulas for ε*_2 and ε*_3 have overlapping, inconsistent cases: each first case is 'ν2>r2' or 'ν3>r2', while the second case is '0 if ν2≥r2' or '0 if ν3≥r2'. These second cases should presumably be 'ν2≤r2' and 'ν3≤r2'.","section":"Section 5, Proposition 5.9(1)"},{"comment":"The notation ν±αi shifts between the weight ν∈−Q+ and the dimension vector (νi)i∈I. This ambiguity contributed directly to the boundary errors in the explicit formulas; the authors should fix a convention and state clearly how a weight ν+αi is translated back to a dimension vector.","section":"Sections 4 and 5"}],"recommendation":"major_revision","confidential_remarks":"The central geometric construction is promising and the main crystal theorem may well survive, but the explicit formulas that support the paper's advertised concrete results are incorrect at boundary weights. The revision should supply corrected operator formulas, re-run the proofs of the corollaries, and fix the internal inconsistency in the proof of Corollary 5.10(2)."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The core idea is new and worth taking seriously: they build Kashiwara crystals on irreducible components of multiparameter persistence module varieties using commutative relations instead of Kashiwara–Saito's preprojective relations, and the general construction in Section 3 is careful. But the explicit 2x2 description, Theorem 5.5, is false as stated. Take the 2x2 grid with ν=(0,1,1,0). The unique component is (ν;0,0). Since vertex 1 has no incoming arrows, ε1=0 on the component, so ~e1 is 0 by Definition 3.11. The theorem instead outputs (ν+α1; −1,0), a \"component\" with negative dimension, which does not exist. This is a concrete counterexample, not a gap in an edge case argument. The theorem needs extra clauses: outputs must be discarded whenever a rank or a dimension becomes negative. As written, the advertised explicit formulas are wrong, and Corollary 5.10 (no embedding into B(∞), non-seminormality, connectivity) relies on those formulas, so those consequences are not established.\n\nWhat is good. The departure from preprojective relations is a real difference, not a cosmetic one: the components of EC(ν) are not equidimensional (Remark 5.2), and the simple pullback/pushforward argument from [16] genuinely fails, so the correspondence in Proposition 3.8 has to do real work. The proof of Proposition 3.8 is detailed, with local trivializations built from gExt_i and the fixed-complement trick; I did not find a hole there. The main theorem 3.13 then follows by a standard crystal-axiom verification, assuming those geometric facts. The parametrization of components in the 2x2 case is not new (Hesselink; Musili–Seshadri), and the authors say so; the crystal operator formulas are the new part, and they are exactly what is broken. The appendix with explicit generic representations via Crawley-Boevey–Schröer is a nice addition.\n\nMinor issues: two clear typos (Theorem 4.1(2) has ν_i≥ν_i, should be ν_i≥ν_{i+1}; Corollary 5.10(2) gives ε1=ε1'=1 but the values from Proposition 5.8 are 1 and 0). The base field is never stated to be algebraically closed, which is standard but should be said. The star-crystal results are stated with proofs omitted.\n\nWho this is for: people in TDA and quiver representations who care about geometric invariants of multiparameter persistence. The main existence theorem looks plausible, but the paper cannot be accepted in this form. It deserves a serious referee, not a desk reject, because the core construction is novel and the flaw is a fixable boundary condition, not a collapse of the central idea. I would ask the authors to correct Theorem 5.5, re-verify Corollary 5.10, and add the missing hypotheses.","headline":"A genuinely new geometric crystal construction on multiparameter persistence varieties, but the explicit 2x2 operator formulas are false as stated and need correction before the paper can be trusted.","tokens_in":28846,"tokens_out":6221,"would_cite":false,"duration_ms":49672,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["16G20","55N31"],"pacs":[],"model":"deepseek-v4-flash","headline":"The set of irreducible components of the varieties of d-parameter persistence modules carries a Kashiwara crystal structure.","keywords":["Multiparameter persistence modules","Kashiwara crystal","Irreducible component","Quiver representations","Commutative grid","Crystal operators","Persistent homology"],"falsifier":"A direct computation for the $2\\times2$ grid at a boundary dimension vector with $\\nu_1+\\nu_4=\\nu_2+\\nu_3$, comparing the closure $\\overline{\\varpi_2\\varpi_1^{-1}(\\bar\\Lambda_1)}$ with the component predicted by Theorem 5.5; if the closure turns out reducible, or if for any vertex $i$ the equality $\\tilde e_i\\tilde f_i\\Lambda=\\Lambda$ fails, the crystal structure would collapse.","tokens_in":27639,"feed_emoji":"💎","tokens_out":11123,"duration_ms":92857,"temperature":0.7,"pith_summary":"This paper establishes that the set $B_C=\\bigsqcup_\\nu B_C(\\nu)$ of all irreducible components of the varieties of $d$-parameter persistence modules carries a Kashiwara crystal structure, with the crystal operators defined geometrically through the variety of extensions of persistence modules. A sympathetic reader should care because multiparameter persistence categories are almost always of infinite representation type, so indecomposable decompositions are infeasible; the crystal graph gives a finite combinatorial skeleton that records how simple pieces can be added or removed one vertex at a time. The construction departs from the earlier geometric route by using the commutative grid quiver itself, with commutative relations, rather than a doubled quiver with preprojective relations, and this forces a careful restriction to open subsets where certain dimensions are minimal. Explicit formulas are worked out for the one-parameter and $2\\times2$ cases, showing the crystal is connected and yet not embeddable into the standard quantum-group crystal.","feed_headline":"Multiparameter persistence components assemble into a crystal","feed_subtitle":"Edges add or remove simple modules; explicit formulas cover the 1- and 2-parameter cases.","key_machinery":"The carrying mechanism is the extension variety $E'_C(\\nu+\\alpha_i,-\\alpha_i)$ with its two projections $\\varpi_1:E'_C\\to E_C(\\nu+\\alpha_i)$ and $\\varpi_2:E'_C\\to E_C(\\nu)$. The Kashiwara operator $\\tilde f_i$ sends a component $\\bar\\Lambda$ to the closure $\\overline{\\varpi_2\\varpi_1^{-1}(\\bar\\Lambda_1)}$, and $\\tilde e_i$ sends $\\Lambda$ to $\\overline{\\varpi_1\\varpi_2^{-1}(\\Lambda_2)}$, where $\\bar\\Lambda_1$ and $\\Lambda_2$ are the open subsets on which $\\dim\\operatorname{gExt}_i$ and $\\varepsilon_i$ take their minimum values. The object $\\operatorname{gExt}_i(f)=\\ker F_i(f)$, assembled from signed sums of maps leaving vertex $i$ in the commutative grid, plays the role of the extension group in earlier constructions, and the proof that these restricted projections are smooth surjections with connected fibers is what makes the image closures irreducible, and hence well-defined components.","core_discovery":"The central claim is that the set $B_C$ of irreducible components of the representation varieties $E_C(\\nu)$ of $d$-parameter persistence modules---representations of the equioriented commutative $d$-grid quiver $\\overrightarrow G_m$---is a Kashiwara crystal with respect to the root system of that quiver. The weight of a component is the negative dimension vector, the maps $\\varepsilon_i$ and $\\varphi_i$ are computed from cokernel dimensions and the Cartan pairing, and the Kashiwara operators $\\tilde e_i,\\tilde f_i$ are defined as closures of the two projections of an extension variety $E'_C(\\nu+\\alpha_i,-\\alpha_i)$; Proposition 3.8 proves these restrictions are smooth surjections with connected fibers, so the closures are irreducible components. This adapts the geometric construction of crystals from quiver representation varieties to commutative relations instead of preprojective relations, and it requires an open-subset adjustment because the component varieties are not equi-dimensional. In the one-parameter case the operators become simple inequalities between adjacent dimension entries, and in the $2\\times2$ case all components are parameterized by rank pairs $(r_1,r_2)$ with explicit operator formulas; the resulting crystal is connected, but it is not upper seminormal and admits no embedding into $B(\\infty)$.","pith_inferences":["Implicit in the proof is a general recipe: whenever an extension variety for a bound quiver with commutative relations admits two smooth surjective projections with connected fibers on open subsets, the same closure construction should yield a crystal; testing this on other finite posets or grid shapes would show how far the method extends.","Because the $2\\times2$ crystal is connected and explicitly computable, one could compare it with rank decompositions or signed barcode invariants of multiparameter persistence and ask whether the crystal graph recovers or refines those invariants in a positive, multiplicity-free way.","The failure of upper seminormality suggests that multiparameter persistence crystals may require a relaxed notion of seminormality; a natural next step is to check whether grids with more than two parameters are connected and whether their components satisfy an analogue of the $\\varepsilon_i=\\varepsilon'_i$ equality away from finitely many dimension vectors.","One could compute the crystal graph for the smallest three-parameter grid and compare the number of components with the $2\\times2$ formulas; a matching pattern would strengthen the conjecture that these component crystals form a systematic family indexed by the shape of the commutative grid."],"forward_implications":["Every $d$-parameter persistence module with a fixed dimension vector has a finite set of irreducible components that can be viewed as vertices of a crystal graph, with edges labeled by adding or removing a simple module at one grid vertex.","In the one-parameter case, the crystal operators reduce to explicit inequalities: $\\tilde e_i$ can act only when $\\nu_{i-1}<\\nu_i$, and $\\tilde f_i$ only when $\\nu_{i-1}\\le\\nu_i$.","In the $2\\times2$ case, the irreducible components are exactly parameterized by rank pairs $(r_1,r_2)$ with $r_1+r_2=\\nu_2+\\nu_3$ when $\\nu_1+\\nu_4\\ge\\nu_2+\\nu_3$, and the crystal graph is connected.","The crystal $B_C$ does not embed into the crystal $B(\\infty)$ of the negative half of the quantum group and is not upper seminormal, so this is a new family of crystals rather than a known quantum-group crystal.","A companion $*$-crystal structure is obtained by dualizing representations and reversing the grid, with corresponding explicit formulas in the one-parameter and $2\\times2$ cases."],"supporting_citations":[{"why":"Defines the equioriented commutative d-grid quiver $\\overrightarrow{G_m}$ whose representations are the multiparameter persistence modules studied throughout.","marker":"[1]"},{"why":"Provides the structure theorem for irreducible components of module varieties, used in the appendix to identify the general representation in each component.","marker":"[6]"},{"why":"Supplies the root system, generalized Cartan matrix, and weight-lattice setup that fix the crystal's simple roots and weights.","marker":"[14]"},{"why":"Defines Kashiwara crystals and the operators $\\tilde e_i, \\tilde f_i$; the axioms the paper verifies are exactly these.","marker":"[15]"},{"why":"The geometric construction of crystals from irreducible components of representation varieties that this paper adapts, replacing preprojective relations by commutative relations.","marker":"[16]"},{"why":"The polyhedral realization of the crystal $B(\\infty)$ used to show $B_C$ cannot embed into $B(\\infty)$.","marker":"[21]"}],"fun_headline_variants":["Kashiwara crystals emerge from multiparameter persistence","Commutative quiver yields crystals for persistence modules","Explicit crystal operators for 1- and 2-parameter persistence","Persistence components assemble into a connected crystal"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction rests on the claim that, on the open regions where the relevant dimensions are minimal, the two projection maps between the extension variety and the representation varieties are surjective with connected, well-behaved fibers; if those maps degenerated anywhere, the closures used to define the crystal operators could fail to be irreducible and the crystal laws could not be checked.","fun_headline_variants_meta":{"raw":{"variants":["Kashiwara crystals emerge from multiparameter persistence","Commutative quiver yields crystals for persistence modules","Explicit crystal operators for 1- and 2-parameter persistence","Persistence components assemble into a connected crystal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000337,"raw_usage":{"total_tokens":1836,"prompt_tokens":891,"completion_tokens":945,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":507,"completion_tokens_details":{"reasoning_tokens":879}},"tokens_in":507,"tokens_out":945,"duration_ms":8359,"temperature":1.0,"reasoning_tokens":879,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T11:41:10.101094+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct computation for the $2\\times2$ grid at a boundary dimension vector with $\\nu_1+\\nu_4=\\nu_2+\\nu_3$, comparing the closure $\\overline{\\varpi_2\\varpi_1^{-1}(\\bar\\Lambda_1)}$ with the component predicted by Theorem 5.5; if the closure turns out reducible, or if for any vertex $i$ the equality $\\tilde e_i\\tilde f_i\\Lambda=\\Lambda$ fails, the crystal structure would collapse.","supporting_citations":[{"cited_title":"Irreducible components of varieties of modules","cited_arxiv_id":null,"evidence_quote":"Provides the structure theorem for irreducible components of module varieties, used in the appendix to identify the general representation in each component."},{"cited_title":"On crystal bases of the Q-analogue of universal enveloping al- gebras","cited_arxiv_id":null,"evidence_quote":"Defines Kashiwara crystals and the operators $\\tilde e_i, \\tilde f_i$; the axioms the paper verifies are exactly these."},{"cited_title":"Geometric construction of crystal bases","cited_arxiv_id":null,"evidence_quote":"The geometric construction of crystals from irreducible components of representation varieties that this paper adapts, replacing preprojective relations by commutative relations."},{"cited_title":"Polyhedral realizations of crystal bases for quantized Kac-Moody algebras","cited_arxiv_id":null,"evidence_quote":"The polyhedral realization of the crystal $B(\\infty)$ used to show $B_C$ cannot embed into $B(\\infty)$."}],"review_version":1}