REVIEW 4 major objections 5 minor 11 references
Combinatorics in (2,1)-categories
T0 review · 4 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash
Pith's one-line read Counting homomorphisms into two relatively finite functors over a locally finite groupoid determines whether they are equivalent.
desk verdict 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. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
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.
What would settle it
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.
Extended reading notes
Core claim
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.
Load-bearing premise
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.
Editorial extensions
If this is right
- 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 $n$.
Reading between the lines
- 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.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
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.
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 (4)
- [Section 5.8, Proposition 5.9] 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.
- [Lemma 6.2] 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.
- [Theorem 6.4 proof, equation (3)] 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.
- [Theorem 3.3] 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.
minor comments (5)
- [Definition 5.3] 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 4] 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.
- [Theorem 6.4 proof, second part] 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.
- [Theorem 6.4 proof, second part] 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.
- [Throughout] There are several typographical errors, including 'isomoprhism' in the introduction and 'ta kes' in the abstract; these should be corrected in a final revision.
Circularity Check
No circular derivation: the main theorem is an external-style Lovász-counting argument, though it contains real proof gaps (unproved orthogonality in Section 5.8 and an invalid fullness inference in Lemma 6.2) that are correctness issues, not circularity.
full rationale
No step in the derivation reduces the conclusion to its own inputs. The main theorem, Theorem 6.4, is proved by adapting Lovász's classical homomorphism-counting argument: Lemma 6.2 decomposes C(X,Y) over E-quotients, induction over a well-founded order on quotients is used to prove equality of faithful-morphism cardinalities, and Lemma 6.3 plus a stabilization argument gives the equivalence. The ingredients are standard external results (Lovász 1967, Baez–Dolan groupoid cardinality, Dupont–Vitale and Kasangian–Vitale factorization systems) and none is a restatement of the theorem. There are no fitted parameters, no data subset is used to predict a closely related quantity, and no load-bearing premise is justified only by a citation to the present author. The genuine weaknesses are rigor gaps, not circularity: Section 5.8 explicitly says the orthogonality/fill-in condition for the lifted ternary factorization system is 'left as an exercise to the reader,' and Lemma 6.2 asserts that a functor that is fully faithful on each coproduct component is 'thus' fully faithful on the entire coproduct, which also requires ruling out 2-morphisms between images of different components. These are unproved mathematical assertions on which the proof depends, but they do not make the argument circular; they make it incomplete. Since the circularity standard requires exhibiting a specific reduction of a claimed prediction or theorem to its inputs, and no such reduction appears, the appropriate score is 0.
Assumptions & free parameters
assumptions (3)
- domain assumption The base groupoid B is locally finite, and relative finiteness is satisfied by the functors involved.
- standard math Groupoid cardinality is well-defined and satisfies the basic properties in Proposition 2.2.
- ad hoc to paper The factorization system on Grpd/B exists and restricts to RelFin_BG.
Cite this review
Pith. "Pith review of Combinatorics in (2,1)-categories." pith.science (2026). https://pith.science/paper/VI2OFMAW
@misc{pith2026250203585,
author = {Pith},
title = {Pith review of: Combinatorics in (2,1)-categories},
year = {2026},
howpublished = {\url{https://pith.science/paper/VI2OFMAW}},
note = {Machine review of arXiv:2502.03585}
}
read the original abstract
Groupoid cardinality is an invariant of locally finite groupoids which has many of the properties of the cardinality of finite sets, but which takes values in all non-negative real numbers, and accounts for the morphisms of a groupoid. Several results on groupoid cardinality are proved, analogous to the relationship between cardinality of finite sets and i.e. injective or surjective functions. We also generalize to a broad class of (2,1)-categories a famous theorem of Lov\'asz which characterizes the isomorphism type of relational structures by counting the number of homomorphisms into them.
Reference graph
Works this paper leans on
-
[1]
John C. Baez and James Dolan. From finite sets to feynman diagrams. 2000. URL https://arxiv.org/abs/math/0004133
arXiv 2000
-
[2]
Lov\'asz Meets Weisfeiler and Leman
Holger Dell, Martin Grohe, and Gaurav Rattan. Weisfeiler-leman meets homomorphisms. CoRR, abs/1802.08876, 2018. URL http://arxiv.org/abs/1802.08876
work page Pith review arXiv 2018
-
[3]
M. Dupont and E.M. Vitale. Proper factorization systems in 2-categories. Journal of Pure and Applied Algebra, 179 0 (1): 0 65--86, 2003. doi:https://doi.org/10.1016/S0022-4049(02)00244-X
-
[4]
S. Kasangian and E.M. Vitale. Factorization systems for symmetric cat-groups. Theory and Applications of Categories, 7: 0 47--70, 2000. URL https://api.semanticscholar.org/CorpusID:122521633
work page 2000
-
[5]
The euler characteristic of a category
Tom Leinster. The euler characteristic of a category. Documenta Mathematica, 2006
work page 2006
-
[6]
L. Lovász. Operations with structures. Acta Mathematica Academiae Scientiarum Hungaricae, 18: 0 321–328, 09 1967. doi:https://doi.org/10.1007/BF02280291
-
[7]
L. Lovász. Direct product in locally finite categories. Acta Scientiarum Mathematicarum, 33, 01 1972
work page 1972
-
[8]
Jacob Lurie. Higher Topos Theory. Annals of Mathematics Studies. Princeton University Press, 2009
work page 2009
Show all 11 references
-
[9]
J. Morton. Categorified algebra and quantum mechanics. Theory and Applications of Categories, 16: 0 785–854, 2006. URL http://www.tac.mta.ca/tac/volumes/16/29/16-29abs.html
2006
-
[10]
Jardine Paul G
John F. Jardine Paul G. Goerss. Simplicial Homotopy Theory. Progress in Mathematics. Birkhäuser Basel, 1999. doi:https://doi.org/10.1007/978-3-0348-8707-6
1999 doi
-
[11]
A. Pultr. Isomorphism types of objects in categories determined by numbers of morphisms. Acta Scientiarum Mathematicarum, 35: 0 155--160, 1973
1973
Reviewed August 9, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.