{"id":"b81711e0-93ad-4e80-9b3e-12f14e7ec479","arxiv_id":"2502.03585","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A generalization of Lovász's homomorphism-counting theorem to a broad class of (2,1)-categories, plus new results on groupoid cardinality.","lead":"This paper proves new properties of groupoid cardinality, a way to count mathematical structures while accounting for their symmetries. Its main result generalizes a famous theorem of Lovász that says a structure is determined by counting the maps into it, extending this to a wide class of higher categories.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 6.4 depends on an unproved ternary factorization system on Grpd/B; the proof of Lemma 6.2 also skips the fullness check across E-quotient components, so the cancellation in equation (3) currently lacks a foundation.","rationale":"The paper's main theorem is a plausible Lovász-style result, and much of the surrounding material (Sections 2-4, tameness of representation groupoids) is independent and seems sound. I am not objecting to the result itself. The load-bearing issue is structural: the proof of Theorem 6.4 is a proof-by-decomposition argument that needs the full machinery of a factorization system on the slice 2-category, and the one part of that machinery that is most used—orthogonality—is explicitly not supplied. The paper's own text marks it as an exercise, which is an admission of missing support rather than a derivation. Lemma 6.2's proof then adds a second skip: global fullness of the functor out of the coproduct is asserted from component-wise fullness, which is not generally valid. The sign error in equation (3) is minor but illustrates that the displayed formal manipulations have not been carefully checked. None of these issues appears to be a fundamental counterexample to the main theorem; the gaps look fillable. For that reason I would keep the reader's conditional verdict, not escalate to rejection. If the fill-in construction turns out to be impossible, the theorem would become unverdictable rather than disproven, because the counting invariant might still determine objects by other means. The reader and I mostly agree; I add the precise failure mode inside Lemma 6.2 and the sign slip.","tokens_in":16590,"tokens_out":30312,"duration_ms":291743,"concrete_test":"Write out a complete proof of the fill-in axiom for the claimed factorization system on Grpd/B from Section 5.8. Concretely, take an arbitrary square with e in E and m in M in Grpd/B, construct the 1-morphism w and the 2-cells α,β, and verify the coherence equation and uniqueness. In parallel, check the missing fullness step of Lemma 6.2: prove that any 2-isomorphism me ⇒ m'e' between composites in different E-quotient components forces the quotients e and e' to be equivalent (via the orthogonality fill-in). If either construction cannot be completed on a test case (e.g. B a non-trivial finite group, X and Y finite groupoids), then equation (3) is not justified and Theorem 6.4 lacks a correct proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central proof of Theorem 6.4 is built on Section 5.8's claim that the ternary factorization system on Grpd lifts to Grpd/B. The author verifies that morphisms admit three-factor decompositions, but the orthogonality/fill-in condition is explicitly left 'as an exercise to the reader' (Section 5.8). This is not a cosmetic omission: Lemma 6.2, used to derive equation (3), assumes a genuine factorization system, and the decomposition of RelFin_BG(S,F) into a coproduct over E-quotients is only meaningful if orthogonality holds. Moreover, the proof of Lemma 6.2 as written is incomplete in a specific way: it says precomposition with representatives is 'fully faithful when restricted to each component, and thus on the entire coproduct'. A functor from a disjoint union is fully faithful only if, in addition to each component map being fully faithful, there are no 2-morphisms in the codomain between images of different components. Ruling this out requires the uniqueness/orthogonality of factorizations, which is exactly what is not proved; without it, distinct E-quotient classes could contribute isomorphic objects to C(X,Y) and the cardinality sum in equation (3) would double-count. The displayed equation (3) also has a sign error (the sum should be subtracted), though this particular error is harmless because the induction makes the summands zero. These gaps are repairable, but until the fill-in is constructed and fullness across components is checked, the proof of the main theorem is unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies groupoid cardinality in (2,1)-categories. It proves formulas and inequalities for groupoid cardinality, computes generating functions for group actions and finite-dimensional representations, constructs relatively finite functor categories, and states a Lovász-type theorem for the (2,1)-category RelFin_B: if |RelFin_B(S,F)| = |RelFin_B(S,F')| for all finite groups H and functors S:BH→B, then F and F' are equivalent. The proof strategy is to use a ternary factorization system on Grpd/B, a coproduct decomposition of hom-groupoids over E-quotients, and a Noetherian induction replacing homomorphism counts by counts of faithful morphisms.","tokens_in":16917,"tokens_out":20690,"duration_ms":201269,"significance":"If the central theorem is fully established, it would be a genuinely interesting higher-categorical analogue of Lovász's theorem, and the framework of relatively finite functors over a locally finite groupoid is a natural setting for combinatorial applications. The paper also contains useful standalone results, such as the inequalities relating fullness/faithfulness to groupoid cardinality and the generating-function computations. A particular strength is that the main argument is not circular: it derives the theorem from standard definitions and prior factorization-system results, with no fitted parameters. However, the main theorem currently rests on an unproved factorization-system claim and an incomplete lemma, so the central proof is not yet complete.","major_comments":[{"comment":"The paper does not prove that the classes (E,~M) and (~E,M) form factorization systems on Grpd/B. It verifies the existence of three-factor decompositions, but the orthogonality/fill-in condition is left to the reader: the text says the first three axioms are 'obvious' and that constructing the natural transformation for the fill-in is 'left as an exercise.' This is load-bearing: Lemma 6.2 and the decomposition used in equation (3) of Theorem 6.4 require a genuine factorization system. Please provide a complete proof of the fill-in condition, including the construction of the natural transformation that makes the fill-in functor a morphism in Grpd/B, the verification of the required 2-cell equation, and the uniqueness of the fill-in up to unique coherent 2-isomorphism, and check that the system restricts to RelFin_BG.","section":"Section 5.8, Proposition 5.9"},{"comment":"The proof that the comparison functor from the coproduct to C(X,Y) is fully faithful is incomplete. The text asserts that the functor is 'fully faithful when restricted to each component, and thus on the entire coproduct,' but this implication requires that there are no 2-morphisms in C(X,Y) between images of morphisms lying in different E-quotient components. Such cross-component 2-morphisms can only be excluded by using the uniqueness/orthogonality of factorizations, e.g. Proposition 2.9; this is not shown. Without this check, the coproduct formula and the cardinality sum in equation (3) could double-count, so the induction in Theorem 6.4 lacks a foundation at this point.","section":"Lemma 6.2"},{"comment":"Equation (3) has a sign error. From Lemma 6.2 one obtains |RelFin(S,F)| = Σ_T |~M(T,F)| and the same for F'. Subtracting gives |~M(S,F)| - |~M(S,F')| = -Σ_{T≠S}(|~M(T,F)| - |~M(T,F')|), not the displayed positive sum. The missing minus sign is harmless in the subsequent induction because the summands are set to zero, but the displayed equation should be corrected.","section":"Theorem 6.4 proof, equation (3)"},{"comment":"The count of the subgroup B_n of block upper triangular matrices is incorrect for dim V > 1. The proof sets #B_n = a^n(a+1)^{T_{n-1}} with a = #Aut(V), but an off-diagonal d×d block over F_q has q^{d^2} possible entries, not a+1 choices; the formula is valid only in the case d = 1. The convergence argument can be repaired by counting block upper triangular matrices with diagonal blocks in Aut(V) and arbitrary off-diagonal blocks, giving #B_n = a^n q^{d^2 n(n-1)/2}, but as written the estimate is unjustified.","section":"Theorem 3.3"}],"minor_comments":[{"comment":"The definition of relative finiteness quantifies over 'every object y of G', but it should quantify over objects of B (the base groupoid), since F^{-1}(y) is defined for y in B.","section":"Definition 5.3"},{"comment":"The notation '/integerdivide' appears in Propositions 4.1 and 4.2 as part of expressions such as 'H/integerdivideφ(G)'; this appears to be a rendering artifact for set-theoretic difference and should be replaced by standard notation.","section":"Section 4"},{"comment":"The claim that the chain of injective homomorphisms 'must eventually stabilize, and thus there is some i for which φ_i is an isomorphism' is not quite correct: the first stabilization could occur at a ψ_i rather than at a φ_i. In that case Lemma 6.3 should be applied to ψ_i, which still yields an equivalence between a component of F and a component of F'. The argument is repairable but the sentence should be revised.","section":"Theorem 6.4 proof, second part"},{"comment":"The final induction proving V ≃ W is not specified. The proof should state the induction measure (for example, the number of connected components of the source groupoids) and verify that V and W satisfy the hypotheses needed to apply the theorem or induction hypothesis after removing the common component U.","section":"Theorem 6.4 proof, second part"},{"comment":"There are several typographical errors, including 'isomoprhism' in the introduction and 'ta kes' in the abstract; these should be corrected in a final revision.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper contains promising ideas and the main theorem may well be true, but the missing factorization-system proof in Section 5.8 is substantial and currently makes the central argument incomplete. The gaps appear repairable within the manuscript's scope, so I recommend major revision rather than rejection. The author should also fix the incorrect count in Theorem 3.3 and the sign error in equation (3)."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main theorem—Lovász-style homomorphism-counting in RelFin_B—is a genuine new contribution, and the proof strategy is reasonable. The paper also has useful results on functor groupoid cardinality and tameness of representation groupoids. I agree with the conditional verdict, though not with every specific worry.\n\nWhat is actually new: the generalization of Lovász to relatively finite functors over a locally finite groupoid seems original. The use of the ternary factorization system on Grpd/B to run the induction is sensible, and the second half of the proof—finding matching connected components via faithful morphisms—is clever. The functor groupoid formulas and the tameness of core(Rep_{f.d.}(G,F_q)) are also new. The count in Theorem 3.3 is fine if you read 'Aut(V)' as the automorphism group of the representation; the off-diagonal blocks are elements of the endomorphism division algebra, so a+1 choices each. Sloppy wording, but correct.\n\nThe real soft spots are in Section 5.8 and Lemma 6.2. The ternary factorization system on Grpd/B is asserted rather than proved; the fill-in condition is explicitly left to the reader, and that condition is load-bearing. Lemma 6.2's proof is incomplete as written: full faithfulness on each component of the coproduct does not imply full faithfulness on the entire coproduct unless there are no 2-morphisms between images of different components. Ruling that out requires exactly the orthogonality that is not proved. Equation (3) also has a sign error, but that one is harmless because the induction kills the sum.\n\nThese gaps look repairable. The paper's architecture is sound, and I expect a knowledgeable referee could fill in the missing diagram chase and fix Lemma 6.2. But as it stands, the main theorem is unsupported at a critical step.\n\nWho this is for: categorists working with groupoid cardinality, stuff types, or higher categorical combinatorics. It deserves serious referee time, but the referee should require the fill-in construction and a correct proof of Lemma 6.2 before acceptance.","headline":"The main theorem is a real generalization, but the proof leans on an unproved factorization system and an incomplete Lemma 6.2; worth refereeing, not accepting yet.","tokens_in":17400,"tokens_out":15870,"would_cite":false,"duration_ms":138781,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["18A99","18A22","18A25","18A32"],"pacs":[],"model":"deepseek-v4-flash","headline":"Counting homomorphisms into two relatively finite functors over a locally finite groupoid determines whether they are equivalent.","keywords":["groupoid cardinality","(2,1)-categories","relatively finite functors","stuff types","ternary factorization systems","homomorphism counting","tame groupoids"],"falsifier":"Construct a 1-morphism in Grpd/B whose lifted (E,~M)-factorization fails the fill-in condition, or produce two non-equivalent relatively finite functors over one locally finite groupoid with identical cardinalities $|\\mathrm{RelFin}_B(S,\\cdot)|$ for all $S \\colon BH \\to B$; either outcome would overturn Theorem 6.4.","tokens_in":16392,"feed_emoji":"🧮","tokens_out":11919,"duration_ms":98465,"temperature":0.7,"pith_summary":"This paper develops groupoid cardinality as a combinatorial invariant in (2,1)-categories and proves a homomorphism-counting theorem in that setting. The main result says that in the category $\\mathrm{RelFin}_B$ of relatively finite functors into a locally finite groupoid $B$, two functors are equivalent if, for every finite group $H$ and every functor $S \\colon BH \\to B$ from the one-object groupoid $BH$, the groupoid of maps from $S$ into each functor has the same groupoid cardinality. The paper also computes cardinalities of functor groupoids via generating functions, proves that the groupoid of finite-dimensional representations of a finite group over a finite field of coprime characteristic is tame, and establishes set-cardinality-style inequalities for full and faithful functors between tame groupoids.","feed_headline":"Counting functors pins down objects in (2,1)-categories","feed_subtitle":"Relatively finite functors over a locally finite groupoid are fixed up to equivalence by hom-counts from finite groups.","key_machinery":"The central object is groupoid cardinality, $|G| = \\sum_{[x]} 1/\\#G_x$, which weights each isomorphism class by the reciprocal of its automorphism group. The load-bearing construction is a ternary factorization system on the slice 2-category $\\mathrm{Grpd}/B$, with classes of essentially surjective full, essentially surjective faithful, and fully faithful functors. Lemma 6.2 uses this factorization system to decompose each hom-groupoid $\\mathrm{RelFin}_B(S,F)$ into a coproduct over $E$-quotients of $S$, turning equality of cardinalities into a well-founded induction; Lemma 6.3 then handles the component-matching step for functors out of deloopings of finite groups.","core_discovery":"The central claim is Theorem 6.4: if $B$ is a locally finite groupoid and $F \\colon G \\to B$ and $F' \\colon G' \\to B$ are relatively finite functors such that $|\\mathrm{RelFin}_B(S,F)| = |\\mathrm{RelFin}_B(S,F')|$ for every finite group $H$ and functor $S \\colon BH \\to B$, then $F$ and $F'$ are equivalent. The proof first reduces to the case $B \\simeq BG$ for a finite group $G$, then uses Noetherian induction over $E$-quotients of $S$ to upgrade equality of hom-counts into equality of counts of faithful morphisms. Relative finiteness then lets the argument split $F$ and $F'$ as $U \\sqcup V$ and $U \\sqcup W$ with $V \\simeq W$, so the two functors are equivalent. Earlier in the paper, groupoid cardinality is shown to behave like set cardinality: full functors satisfy $|G| \\le |H|$, essentially surjective faithful functors satisfy $|G| \\ge |H|$, and under equal total cardinality these conditions force a functor to be an equivalence.","pith_inferences":["One consequence not stated in the paper is that the same finite-group hom-counts could serve as a practical isomorphism test: if two relatively finite functors agree on all counts from functors $BH \\to B$, no finite probing by such functors can distinguish them, so the counts are a complete invariant in principle.","The proof suggests a general recipe for other (2,1)-categories: any category with a well-founded ternary factorization system whose first class is fully cofaithful should admit the same hom-counting theorem, even though the paper does not isolate such axioms.","If Conjecture 3.4 holds, the tameness of modular representation groupoids would let the same counting invariants count representations over fields whose characteristic divides the group order, extending the semisimple case proved here.","The failure of the component-matching argument for 2-groupoids, noted in Section 7, indicates that an $\\infty$-categorical analogue of the theorem, if it exists, will require new ideas beyond Postnikov truncation."],"forward_implications":["If Theorem 6.4 is correct, finite-group hom-counts form a complete invariant for relatively finite functors over a locally finite groupoid: two such functors are equivalent exactly when all these counts agree.","The proof gives a cancellation principle: whenever $F \\simeq U \\sqcup V$ and $F' \\simeq U \\sqcup W$ are relatively finite functors with matching hom-counts, the leftover pieces $V$ and $W$ are equivalent.","For the groupoid of finite sets and bijections, the theorem specializes to relatively finite stuff types, and the generating-function formulas of Section 3 make some of the relevant cardinalities computable in closed form.","The Section 4 results give groupoid-cardinality analogues of set-cardinality facts: full functors do not increase cardinality, essentially surjective faithful functors do not decrease it, and equal cardinality turns these conditions into equivalences.","The Postnikov inequalities of Section 7 show that homotopy cardinality of $\\infty$-groupoids is ordered by connectivity and truncation: $|\\mathrm{im}_n f| \\ge |\\mathrm{im}_{n-1} f|$ for even $n$ and $\\le$ for odd $n$."],"supporting_citations":[{"why":"Supplies the original homomorphism-counting theorem and the induction-via-factorization proof that Theorem 6.4 generalizes.","marker":"Lovász [1967]"},{"why":"Introduces groupoid cardinality, the invariant used throughout the paper.","marker":"Baez and Dolan [2000]"},{"why":"Provides the definition of factorization systems in 2-categories and the uniqueness of factorizations used in Section 2.7.","marker":"Kasangian and Vitale [2000]"},{"why":"Defines fully faithful and fully cofaithful 1-morphisms and proper factorization systems, used to lift the factorization to Grpd/B.","marker":"Dupont and Vitale [2003]"},{"why":"Defines stuff types, the motivating class of objects in RelFin_B and the source of the generating-function computations.","marker":"Morton [2006]"}],"fun_headline_variants":["Hom-counts from finite groups fix functors up to equivalence","Counting morphisms from finite groups decides functor equivalence","Categorified Lovász: hom-counts characterize relative finite functors","Finite group hom-counts determine relative finite functors"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes the standard factorization of groupoid functors lifts to Grpd/B as a ternary factorization system satisfying the fill-in condition; the fill-in construction is left to the reader, and the decomposition in Lemma 6.2 depends on it.","fun_headline_variants_meta":{"raw":{"variants":["Hom-counts from finite groups fix functors up to equivalence","Counting morphisms from finite groups decides functor equivalence","Categorified Lovász: hom-counts characterize relative finite functors","Finite group hom-counts determine relative finite functors"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000183,"raw_usage":{"total_tokens":1291,"prompt_tokens":901,"completion_tokens":390,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":517,"completion_tokens_details":{"reasoning_tokens":320}},"tokens_in":517,"tokens_out":390,"duration_ms":4062,"temperature":1.0,"reasoning_tokens":320,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T04:27:44.785952+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a 1-morphism in Grpd/B whose lifted (E,~M)-factorization fails the fill-in condition, or produce two non-equivalent relatively finite functors over one locally finite groupoid with identical cardinalities $|\\mathrm{RelFin}_B(S,\\cdot)|$ for all $S \\colon BH \\to B$; either outcome would overturn Theorem 6.4.","supporting_citations":[],"review_version":1}