Pith. sign in

REVIEW 4 major objections 4 minor 29 references

Sheaves on a bicategory

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves an adjunction making complete enriched categories into the sheaves on a bicategory, unifying quantaloid and monoidal sheaf theory.

desk verdict A genuinely unifying framework that is currently ill-typed as stated: P relies on unique representatives that completeness alone does not guarantee, and the proof leaves too many coherence checks to the reader. read the letter →

arxiv 2507.20820 v1 pith:MYYZ2DGM submitted 2025-07-28 math.CT

Olivia Caramello , Elio Pivet This is my paper · ORCID
classification math.CT MSC 18D2018F1018D3018N10
keywords bicategoricalenrichmentCauchy-completeenrichedcategoriessheavesonbicategoriesGrothendieckconstructionquantaloidstoposespseudofunctorsadjunctions
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper develops a theory of categories enriched in a bicategory $\mathcal{B}$ and proves that complete such categories stand in an adjunction with pseudofunctors $\mathrm{Map}(\mathcal{B})^{\mathrm{coop}} \to \mathbf{Cat}$. The construction is meant to unify two existing strands: sheaves over quantaloids and enrichment over monoidal categories. If the adjunction is correct, the category of complete $\mathcal{B}$-categories deserves to be called the category of sheaves on $\mathcal{B}$, giving a candidate notion of enriched topos. The proof works through an enriched Grothendieck construction $\int$, a fiber pseudofunctor $P$, and a completion functor $C$ built from singletons. In the quantaloid and monoidal special cases the paper recovers the known sheaf-theoretic results and characterizes the fixed points of the adjunction.

What carries the argument

The load-bearing objects are the singletons of a $\mathcal{B}$-category: distributors $\sigma : (\mathrm{id}_*,*) \to (M,A)$ that have right adjoints, with representable singletons $M(-,a)$ as the basic examples. Completeness says every singleton is representable. The construction $\int F$ builds a $\mathcal{B}$-category whose hom-cells are colimits indexed by maps $f$ equipped with arrows $u : b \to F(f)(a)$, giving an enriched Grothendieck construction; the functor $P$ sends a complete $\mathcal{B}$-category to the pseudofunctor of its fibers, using representability of $M(-,a)\gamma$ to define transition functors along maps $\gamma$. The completion functor $C$ sends a $\mathcal{B}$-category to its category of singletons. The adjunction is assembled from these three pieces.

What would settle it

Take $\mathcal{B} = \mathcal{B}_{\mathbf{Set}}$ and the complete $\mathcal{B}$-category with two objects $a_1,a_2$ and $M(a_i,a_j)=\{*\}$ for all $i,j$, the chaotic category equivalent to the terminal category. Then $M(-,a_1)=M(-,a_2)$ as singletons, so for the identity map $\gamma = \mathrm{id}$, the element $\gamma\cdot a_1$ is not unique: both $a_1$ and $a_2$ represent the same singleton, and $P(\mathrm{id})$ is not a well-defined functor. Checking this example settles whether $P$ is defined on all complete $\mathcal{B}$-categories or only on skeletal ones.

Watch

Extended reading notes

Core claim

The paper's central claim is Theorem 3.3.1: for a locally cocomplete closed bicategory $\mathcal{B}$, there is an adjunction $C\int \dashv P$ between the category $\mathbf{Cat}_\kappa(\mathcal{B})$ of complete $\mathcal{B}$-categories and the category of oplax-natural transformations between pseudofunctors $\mathrm{Map}(\mathcal{B})^{\mathrm{coop}} \to \mathbf{Cat}$. Here $\int$ is a Grothendieck construction turning an indexed family of categories into a $\mathcal{B}$-category, and $P$ assigns to a complete $\mathcal{B}$-category its system of fiber categories, whose transition functors are induced by representability of singletons. The paper interprets this as saying that complete $\mathcal{B}$-categories are the sheaves on $\mathcal{B}$, and spells out special cases: for quantaloids the adjunction restricts to a left-exact reflection yielding set-valued sheaves on sites, and for monoidal categories it recovers Cauchy completion and the underlying-category functor.

Load-bearing premise

The load-bearing premise is that completeness makes the representatives $\gamma\cdot a$ unique, not merely existent; without that, the fiber pseudofunctor $P$ is not well defined on all of $\mathbf{Cat}_\kappa(\mathcal{B})$.

Editorial extensions

If this is right

  • If the adjunction is correct, $\mathbf{Cat}_\kappa(\mathcal{B})$ can be treated as the category of sheaves on $\mathcal{B}$, and in well-behaved cases it should inherit the good categorical properties expected of an enriched topos.
  • In the quantaloid case, the adjunction restricts to a reflection $\mathbf{Cat}^{\sigma}_\kappa(Q) \to [\mathrm{Map}(Q)^{\mathrm{op}}, \mathbf{Set}]$; for Grothendieck quantaloids the reflection is left-exact, recovering the site-theoretic sheaf results.
  • For a locale $X$, symmetric complete $R(X)$-categories are exactly sheaves on $X$, recovering Walters' characterization.
  • For a monoidal category $\mathcal{V}$, the functor $P_{-}(*)$ is the underlying-category functor $\mathcal{V}\text{-}\mathbf{Cat}(I,-)$, and when every object of $\mathcal{V}$ is a colimit of copies of $I$ (as for $\mathbf{Set}$ and $\mathbf{Ab}$), every complete $\mathcal{V}$-category is a fixed point of the adjunction.
  • When $P$ is fully faithful, $\mathbf{Cat}_\kappa(\mathcal{B})$ becomes a reflexive subcategory of the 2-presheaf category, making completion behave like sheafification.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper's definition of $P$ uses 'the' representative $\gamma \cdot a$ of $M(-,a)\gamma$. Since completeness alone guarantees existence but not uniqueness, a robust version of the result would either restrict $\mathbf{Cat}_\kappa(\mathcal{B})$ to skeletal objects or move to a 2-categorical formulation where representatives are chosen up to coherent isomorphism.
  • The denseness condition noted in the paper (every 1-cell a colimit of maps) is what upgrades the adjunction to pseudonatural transformations and makes $P$ full. Locating further bicategories satisfying it, beyond $\mathbf{Set}$ and $\mathbf{Ab}$, would give a concrete test of how general the enriched-topos interpretation is.
  • The paper leaves the symmetrization of the Grothendieck construction open in general; its suggestion of a lax-idempotent monad could be developed, and doing so would likely give the symmetric adjunction needed for set-valued sheaves in bicategories that are not quantaloids.
  • A natural testable extension is to check left-exactness of the reflection directly on finite bilimits in $\mathbf{Cat}_\kappa(\mathcal{B})$ for concrete bicategories such as $\mathcal{V} = \mathbf{Cat}$ or simplicial enrichment, following the paper's pointer that left-exactness must be verified on all finite bilimits.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper develops a theory of enrichment over a bicategory B, defines complete B-categories as a generalization of Cauchy-complete enriched categories, and proposes an enriched Grothendieck construction ∫ together with a fiber pseudofunctor P. Its main claim (Theorem 3.3.1) is an adjunction C∫ ⊣ P between complete B-categories and oplax-natural transformations of pseudofunctors Map(B)^coop → Cat. The paper further claims recoveries of Walters' sheaf-on-quantaloid results and discusses the monoidal case, including fixed points of the adjunction for Set and Ab.

Significance. If the main theorem can be made correct, the framework is genuinely unifying: it places quantaloid-enriched and monoidal-enriched category theory under one bicategorical umbrella, and it offers a concrete candidate for a notion of enriched topos. The paper's recoveries of known results about sheaves on locales and sites (Examples 4.1.15 and 4.1.16) and its detailed treatment of the monoidal case give strong evidence that the intended constructions are on the right track. The exposition of Street's enrichment theory is also useful and generally careful. However, the central construction P is currently not well-defined on the stated category, and several load-bearing verifications in the proof of the main adjunction are explicitly deferred to the reader; the advertised significance is therefore not yet realized.

major comments (4)
  1. [§3.2, Definition 3.2.1 and Lemma 3.2.2] The definition of P is not well-defined on the stated domain. For a map γ : x1 → x2, PM,A(γ)(a) is declared to be "the unique element γ·a" representing the singleton M(−,a)γ. But completeness, as defined in Definition 2.3.24, is stated only for skeletal B-categories and gives only existence of a representing object, not uniqueness. Uniqueness is exactly the skeletal condition from Definition 2.3.20, yet Catκ(B) in Definition 2.3.30 is the full subcategory of Cat(B) on all complete B-categories, with no skeletal restriction. A concrete failure is given by B = BSet and the B-category with A = {a,b} and all hom-sets equal to a singleton: this category is Karoubi complete but not skeletal, and M(−,a) = M(−,b) even though a ≠ b, so for γ = id the singleton M(−,a)γ has two representatives. Hence PM,A(id) is not a function, and Theorem 3.3.1 is ill-typed as stated.
  2. [§2.3, Proposition 2.3.33 and Corollary 2.3.34] The same uniqueness problem affects the counit of the Cauchy-completion adjunction. The counit ε(M,A) sends a singleton σ = M(−,a) to "the unique element representing it", and the proof of the triangle identities is explicitly restricted to "any skeletal complete B-category". Corollary 2.3.34, which says that completion of an already complete category yields the same category, also relies on this uniqueness. Thus the adjunction C ⊣ i is established only for skeletal complete B-categories, which does not match the definition of Catκ(B) used in the main theorem.
  3. [§3.3, Theorem 3.3.1 proof and Proposition 3.2.3] Several load-bearing parts of the proof are left to the reader. The text states: "We do not prove the coherence conditions which ensure that A(g) is an oplax-natural transformation", "We leave to the reader the proof that this indeed commutes well with the base of the colimit", and "we do not write the proof that G(α) is a B-functor". Proposition 3.2.3 similarly concludes with "We do not give here a proof of this oplax-naturality". These are not peripheral coherence computations: they are exactly what makes A and G well-defined maps between the hom-sets of the claimed adjunction. Moreover, Theorem 3.3.1 asserts pseudonaturality of the bijection in both variables, but no verification of that pseudonaturality is supplied. The main theorem is therefore not proved as written.
  4. [§2.3, Example 2.3.31 and §4.2, Example 4.2.3] The advertised recoveries are incompatible with a skeletal-only formulation. Example 2.3.31(1) states that Catσκ(BSet) is the category of all small Karoubi-complete categories, and Example 4.2.3 says the adjunction becomes the usual adjunction between categories and Karoubi-complete categories. But ordinary Karoubi-complete categories need not be skeletal; the two-object non-skeletal category described in the first major comment is a counterexample. If the paper is revised by restricting Catκ(B) to skeletal complete B-categories, these examples must be changed accordingly, and the claimed recovery of the usual Karoubi-completion adjunction would no longer be literal.
minor comments (4)
  1. [§1, Introduction] The word "eludication" on page 2 should be "elucidation".
  2. [§4.1, Proposition 4.1.9] In the proof, after showing that every h in the join defining M(−,a)f also appears as hf in the join defining M(−,F(f)(a)), one obtains M(−,a)f ≤ M(−,F(f)(a)). The displayed inequality "M(−,F(f)(a)) ≤ M(−,a)f" therefore has the wrong direction, and the conclusion needs a separate equality argument.
  3. [§3.2, Definition 3.2.1] Definition 3.2.1 assumes B is an involutive bicategory, but Theorem 3.3.1 and Section 3.1 do not state this hypothesis. Please clarify whether the main adjunction really requires involutivity, or whether it is needed only for the symmetric variant discussed in Section 4.1.
  4. [§3.2, proof of Definition 3.2.1] In the displayed computation proving functoriality of PM,A(γ), the expression "M(c·g,b)" appears to be a typo for "M(c·γ,b)"; please correct it.

Circularity Check

0 steps flagged · score 2.0 of 10

Main adjunction is self-contained; minor self-citation (Street [21], Heymans [9]) is ancillary, and the unique-representative assumption in Definition 3.2.1 is a stated gap that weakens the main theorem but does not make it circular by construction.

full rationale

The paper's central claim, the adjunction C∫ ⊣ P (Theorem 3.3.1), is proved by explicit construction: ∫ is built as a colimit-enriched Grothendieck construction (Definition 3.1.1) and P is defined fiberwise (Definition 3.2.1), with the adjunction maps A and G spelled out and verified componentwise in the proof of Theorem 3.3.1. The proof does not invoke the conclusion as an input, and the external results cited (Street [21] for the general theory of enrichment over bicategories, Walters [25,26] and Heymans [9] for the quantaloid recovery) are used as benchmarks or background rather than as the force establishing the adjunction. The 'reader's take' correctly identifies a genuine gap: Definition 3.2.1 defines γ·a as 'the unique element' representing the singleton M(−,a)γ, while Definition 2.3.24 defines completeness only for skeletal B-categories and provides existence, not uniqueness; Definition 2.3.30 then defines Catκ(B) without a skeletal restriction, so P is not well-defined on all objects of its stated domain. This is a correctness/well-definedness defect, not a circular step: it does not make the adjunction equivalent to its own input by construction. The self-citation pattern is moderate: the paper relies technically on Street [21] and recovers Heymans [9] in Section 4, but both are external authors (no author overlap), so they are genuine external support rather than load-bearing self-citation. Accordingly the circularity score is low, reflecting one noticeable definitional gap adjacent to the central construction but no reduction of the result to a fitted or self-cited assumption.

Assumptions & free parameters 0 free parameters · 4 assumptions · 3 invented entities

The central claim rests on standard domain assumptions about the ambient bicategory (local cocompleteness, closedness, optional involution) and on the completeness of V in the monoidal case. There are no fitted parameters. The new definitions (complete B-categories, P, ∫) are mathematical constructions with external benchmarks, so they are listed with independent evidence rather than being ad hoc postulates.

assumptions (4)
  • domain assumption B is locally cocomplete (all hom-categories have all colimits) and closed (right Kan extensions and lifts exist).
    Stated at the start of Section 2; needed so pre- and post-composition functors preserve colimits, which is used to define distributor composition via coends and to define the Grothendieck construction as a colimit.
  • domain assumption When symmetric versions are considered, B is equipped with an involution (−)°: B^op → B that is the identity on objects.
    Assumed in Section 2 for symmetric B-categories (Definition 2.1.19) and used throughout Section 4 for the symmetric cases, including the recovery of Set-valued sheaves.
  • domain assumption For the monoidal examples, V is a Bénabou cosmos (symmetric monoidal closed, complete and cocomplete).
    Stated in Section 4.2; the monoidal fixed point results rely on V's completeness and on the condition that every object is a colimit of copies of I (Proposition 4.2.4).
  • standard math The axiom of choice is used to identify posetal groupoids with sets in Proposition 4.1.3.
    The text says 'we assume the axiom of choice, which is necessary to construct the inverse image of a connected component'; this is a standard set-theoretic assumption, not specific to the paper.
invented entities (3)
  • Complete B-categories (Catκ(B)) independent evidence
    purpose: The 'sheaves' of the theory: B-categories in which every right-adjoint distributor is representable.
    The definition is new in this generality and is validated by recovering the known categories of sheaves on a locale and Karoubi-complete categories for V=Set (Examples 2.3.31).
  • The fiber pseudofunctor P_M,A: Map(B)^coop → Cat independent evidence
    purpose: Assigns to each complete B-category its family of 'fiber' categories, forming the right adjoint in the main adjunction.
    When B is a quantaloid, P recovers the presheaf functor on Map(Q); the recovery of Walters' theorem (Example 4.1.15) provides external validation.
  • The enriched Grothendieck construction ∫ independent evidence
    purpose: Builds a B-category from a pseudofunctor Map(B)^coop → Cat, giving the left adjoint in the main adjunction.
    Specializes to the ordinary Grothendieck construction for B=Set and to the free V-category construction in the monoidal case (Proposition 4.2.5).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sheaves on a bicategory." pith.science (2026). https://pith.science/paper/MYYZ2DGM

@misc{pith2026250720820,
  author       = {Pith},
  title        = {Pith review of: Sheaves on a bicategory},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MYYZ2DGM}},
  note         = {Machine review of arXiv:2507.20820}
}
read the original abstract

We give a detailed account of the theory of enrichment over a bicategory and show that it establishes a two-fold generalization of enrichment over both quantaloids and monoidal categories. We define complete B-categories, a generalization of Cauchy-complete enriched categories serving as a basis for the development of sheaf theory in the enriched setting. We prove an adjunction between complete B-categories and 2-presheaves on the category Map(B) of left adjoints in B. We express conditions under which this adjunction becomes a left-exact reflection, yielding back the usual results linking sheaves on sites and enriched categories. We prove that our adjunction recovers the already existing results about quantaloids, and discuss the fixed points of the adjunction in the monoidal case.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 25 canonical work pages

  1. [1]

    A theory of quantal sets

    Ugo BerniCanani, Francis Borceux, and Rosanna SucciCruciani. “A theory of quantal sets”. In:Journal of Pure and Applied Algebra62.2 (1989), pp. 123–136

    work page 1989

  2. [2]

    Handbook of categorical algebra

    Francis Borceux. Handbook of categorical algebra. 2: Categories and structures. English. Paperback reprint of the hardback edition 1994. Vol. 51. Encycl. Math. Appl. Cambridge: Cambridge University Press,

    work page 1994

  3. [3]

    Quantales and their sheaves

    Francis Borceux and Gilberte van den Bossche. “Quantales and their sheaves”. English. In:Order 3 (1986), pp. 61–87.issn: 0167-8094.doi: 10.1007/BF00403411

    work page doi:10.1007/bf00403411 1986

  4. [4]

    Cauchy completion in cate- gory theory

    Francis Borceux and Dominique Dejean. “Cauchy completion in cate- gory theory”. English. In:Cah. Topologie Géom. Différ. Catégoriques 27.2 (1986), pp. 133–146.issn: 0008-0004. url: https://eudml.org/ doc/91378

    work page 1986

  5. [5]

    Atheoryofenrichedsheaves

    FrancisBorceuxandCarmenQuinteiro.“Atheoryofenrichedsheaves”. English. In: Cah. Topologie Géom. Différ. Catégoriques37.2 (1996), pp. 145–162. issn: 0008-0004

    work page 1996

  6. [6]

    Olivia Caramello and Riccardo Zanfa.Relative topos theory via stacks

    work page

  7. [7]

    Quantal sets and sheaves over quantales

    RP Gylys. “Quantal sets and sheaves over quantales”. In:Lithuanian Mathematical Journal34.1 (1994), pp. 8–29

    work page 1994

  8. [8]

    Sheaves on quantaloids

    RP Gylys. “Sheaves on quantaloids”. In: Lithuanian Mathematical Journal 40.2 (2000), pp. 105–134

    work page 2000

Show all 29 references
  1. [9]

    Sheaves on Quantales as Generalized Metric Spaces

    Hans Heymans. “Sheaves on Quantales as Generalized Metric Spaces”. PhD thesis. Universiteit Antwerpen, 2010

    2010

  2. [10]

    Sheaves on involutive quantales : Grothendieck quan- tales

    Hans Heymans. “Sheaves on involutive quantales : Grothendieck quan- tales”. English. In:Fuzzy Sets and Systems256 (2014), pp. 117–148

    2014

  3. [11]

    Symmetry and Cauchy completion of quantaloid-enriched categories

    Hans Heymans and Isar Stubbe. “Symmetry and Cauchy completion of quantaloid-enriched categories”. English. In:Theory Appl. Categ. 25 (2011), pp. 276–294.issn: 1201-561X. url: https://eudml.org/ doc/225546

    2011

  4. [12]

    Niles Johnson and Donald Yau.2-dimensional categories. English. Ox- ford: Oxford University Press, 2021.isbn: 978-0-19-887137-8; 978-0- 19-887138-5. doi: 10.1093/oso/9780198871378.001.0001

    2021

  5. [13]

    Basic concepts of enriched category theory

    G. M. Kelly. “Basic concepts of enriched category theory”. English. In: Repr. Theory Appl. Categ.2005.10 (2005), pp. 1–136. 71

    2005

  6. [14]

    Flat vs. filtered colimits in the enriched context

    Stephen Lack and Giacomo Tendas. “Flat vs. filtered colimits in the enriched context”. English. In:Adv. Math.404 (2022). Id/No 108381, p. 50. issn: 0001-8708. doi: 10.1016/j.aim.2022.108381

    2022

  7. [15]

    Metric spaces, generalized logic, and closed categories

    F. William Lawvere. “Metric spaces, generalized logic, and closed categories.” English. In: Rend. Semin. Mat. Fis. Milano 43 (1974), pp. 135–166. issn: 0370-7377. doi: 10.1007/BF02924844

    1974 doi

  8. [16]

    Christopher J. Mulvey. “&”. English. In: Suppl. Rend. Circ. Mat. Palermo (2)12 (1986), pp. 99–104.issn: 1592-9531

    1986

  9. [17]

    Quantales: Quantal sets

    Mohammad Nawaz. “Quantales: Quantal sets.” PhD thesis. University of Sussex, 1985

    1985

  10. [18]

    Cauchy complete category

    nLab authors. Cauchy complete category. https : / / ncatlab . org / nlab/show/Cauchy+complete+category. Revision 73. May 2025

    2025

  11. [19]

    Rosenthal

    Kimmo I. Rosenthal. The theory of quantaloids. English. Harlow: Ad- dison Wesley Longman, 1996.isbn: 0-582-29440-1

    1996

  12. [20]

    Characterization of bicategories of stacks

    Ross Street. Characterization of bicategories of stacks. English. Cat- egory theory. Applications to algebra, logic and topology, Proc. int. Conf., Gummersbach 1981, Lect. Notes Math. 962, 282-291 (1982). 1982

    1982

  13. [21]

    Enriched categories and cohomology

    Ross Street. “Enriched categories and cohomology”. English. In:Repr. Theory Appl. Categ.2005.14 (2005), pp. 1–18.url: www.tac.mta.ca/ tac/reprints/articles/14/tr14abs.html

    2005

  14. [22]

    Categorical structures enriched in a quantaloid: cate- gories, distributors and functors

    Isar Stubbe. “Categorical structures enriched in a quantaloid: cate- gories, distributors and functors”. English. In:Theory Appl. Categ.14 (2005), pp. 1–45.issn: 1201-561X

    2005

  15. [23]

    Introducing sheaves over commutative semicartesian quantales

    Ana Luiza Tenório, Caio de Andrade Mendes, and Hugo Luiz Mariano. “Introducing sheaves over commutative semicartesian quantales”. In: arXiv preprint arXiv:2204.08351(2022)

    2022 arXiv

  16. [24]

    Quantaloids and non-commutative ring representations

    Gilberte Van den Bossche. “Quantaloids and non-commutative ring representations”. In:Applied Categorical Structures3 (1995), pp. 305– 320

    1995

  17. [25]

    Sheaves and Cauchy-complete categories

    R. F. C. Walters. “Sheaves and Cauchy-complete categories”. English. In: Cah. Topologie Géom. Différ. Catégoriques22 (1981), pp. 283–286. issn: 0008-0004

    1981

  18. [26]

    Sheaves on sites as Cauchy-complete categories

    R. F. C. Walters. “Sheaves on sites as Cauchy-complete categories”. English. In:J. Pure Appl. Algebra24 (1982), pp. 95–102.issn: 0022-

    1982

  19. [2008]

    isbn: 978-0-521-06122-3

  20. [2021]

    arXiv: 2107.04417 [math.AG]

    arXiv

  21. [4049]

    72 Olivia Caramello Dipartimento di Scienza e Alta Tecnologia, Università degli Studi dell’Insubria, via V alleggio 11, 22100 Como, Italy

    doi: 10.1016/0022-4049(82)90061-5. 72 Olivia Caramello Dipartimento di Scienza e Alta Tecnologia, Università degli Studi dell’Insubria, via V alleggio 11, 22100 Como, Italy. E-mail address: olivia.caramello@uninsubria.it Istituto Grothendieck ETS, Corso Statuto 24, 12084 Mondo...

    doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.