Pith. sign in

REVIEW 3 major objections 6 minor 13 references

On a (terminally connected, pro-etale) factorization of geometric morphisms

T0 review · 3 major / 6 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read Every geometric morphism between Grothendieck topoi factors, essentially uniquely, as a terminally connected morphism followed by a pro-etale morphism—a decomposition that reduces to the classical connected–etale factorization in the…

desk verdict A strong candidate for a canonical factorization of all geometric morphisms, but the proof of Theorem 3.2.1 has a load-bearing gap in the terminal connectedness argument. read the letter →

arxiv 2502.04213 v1 pith:WKBSFO5P submitted 2025-02-06 math.CT

Olivia Caramello , Axel Osmond This is my paper · ORCID
classification math.CT MSC 18B2518A3218N10
keywords geometricmorphismpro-etaleterminallyconnectedfactorizationsystemGrothendiecktoposcofilteredbilimitglobalelementspro-discreteinternallocale
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 proves that every geometric morphism between Grothendieck topoi admits an essentially unique factorization into a terminally connected morphism followed by a pro-etale morphism. The first factor is the best one-sided approximation to 'connected components' that exists without local connectedness; the second factor is a cofiltered limit of etale slice topoi, hence a pro-discrete object over the base. The value of the theorem is that a decomposition once reserved for locally connected or essential morphisms now exists for all geometric morphisms, and it specializes to the classical (connected, etale) factorization in the locally connected case. The construction is canonical: it is indexed by all global elements of the inverse image functor, so no choices of sites or presentations are involved.

What carries the argument

The load-bearing objects are the category $1_{\mathcal{F}} \downarrow f^*$ of global elements of the inverse image functor, the cofiltered bilimit $\mathrm{bilim}_{1_{\mathcal{F}} \downarrow f^*} \mathcal{E}/E$, and the correspondence between global elements of $f^*$ and factorizations of $f$ through etale morphisms. The category of global elements is cofiltered, and an accessibility argument (Lemma 3.1.2) shows it contains a small cofinal subcategory, so the bilimit is a well-defined Grothendieck topos via the cited construction of 2-cofiltered bilimits of topoi. The proof displays all global elements of $f^*$ in the bilimit, which forces the residual map to reflect global elements uniquely; that uniqueness is exactly terminal connectedness.

What would settle it

A concrete way to test the theorem is to compute the canonical map $\mathcal{F} \to \mathrm{bilim}_{1_{\mathcal{F}} \downarrow f^*} \mathcal{E}/E$ for a known geometric morphism and check whether it lifts global elements uniquely; if any global element failed to lift, terminal connectedness would fail. Equivalently, one could seek a pro-etale geometric morphism that is essential but not etale, since Proposition 4.1.1 proves no such morphism exists.

Watch

Extended reading notes

Core claim

The central claim, Theorem 3.2.1, is that for any geometric morphism $f: \mathcal{F} \to \mathcal{E}$ between Grothendieck topoi there is an essentially unique factorization $f = p \circ t$ with $t$ terminally connected and $p$ pro-etale. Terminal connectedness means the inverse image $t^*$ lifts global elements uniquely, i.e. it induces an isomorphism $\mathcal{F}[1_{\mathcal{F}}, t^*(-)] \simeq \mathcal{E}[1_{\mathcal{E}}, -]$; pro-etale means $p$ is equivalent over $\mathcal{E}$ to a cofiltered bilimit of etale morphisms $\mathcal{E}/E \to \mathcal{E}$. The paper constructs the pro-etale factor as the cofiltered bilimit $\mathrm{bilim}_{1_{\mathcal{F}} \downarrow f^*} \mathcal{E}/E$ indexed by all global elements of $f^*$, and shows the comparison map from $\mathcal{F}$ into this bilimit lifts global elements uniquely, so the left factor is terminally connected. This gives a canonical decomposition that reduces to the known (terminally connected, etale) factorization of essential morphisms and, further, to the classical (connected, etale) factorization of locally connected morphisms.

Load-bearing premise

The proof requires that the large category of all global elements of the inverse image $f^*$ contains a small cofinal subcategory, so that the cofiltered bilimit used to build the pro-etale factor is a well-defined Grothendieck topos; this is obtained by choosing a cardinal $\mu$ for which the terminal object is $\mu$-compact and $\mu$-compact objects generate the topos.

Editorial extensions

If this is right

  • Every geometric morphism between Grothendieck topoi now carries a canonical two-stage decomposition, so the connected–etale picture extends beyond locally connected and essential morphisms.
  • Pro-etale morphisms are localic and correspond to pro-discrete internal locales, so the 'connected components' of an arbitrary geometric morphism exist as a pro-discrete locale even when no object of connected components exists.
  • Pro-etale morphisms are discrete opfibrations at the level of points: a point of the pro-etale factor is a coherent family of points of the base together with descent data in the fibers.
  • Terminally connected morphisms are exactly the left orthogonal class to pro-etale morphisms, and they are stable under pullback along tidy morphisms and under bicomma squares.
  • The factorization is oplax functorial: every geometric transformation between two geometric morphisms induces a canonical transformation between their pro-etale middle terms.

Reading between the lines

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

  • The factorization appears to be the topos-theoretic counterpart of the comprehensive (initial, discrete opfibration) factorization of functors; the paper leaves open a precise 'topological initialness' for terminally connected morphisms, and that would be a testable refinement.
  • One could test the construction on morphisms of schemes or presheaf topoi: the pro-etale factor should recover the germ and cofiltered-intersection construction at a point, suggesting a concrete bridge to Grothendieck–Verdier localization.
  • The cardinal-sensitive lemmas (3.1.2 and 5.2.3) suggest that the factorization is presentation-independent up to equivalence but may have a minimal cardinal of definition; computing that cardinal for concrete topoi would be a natural extension.
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

3 major / 6 minor

Summary. The paper extends the classical (connected, étale) factorization of locally connected geometric morphisms to all geometric morphisms between Grothendieck topoi. For any f : F → E, it defines B as the cofiltered bilimit of étale morphisms indexed by the category 1_F ↓ f^* of global elements of f^*, and claims a factorization f = π_f ∘ l_f where l_f : F → B is terminally connected and π_f : B → E is pro-étale. The paper also introduces pro-étale geometric morphisms as cofiltered bilimits of étale morphisms, gives an intrinsic characterization of them via generation under fibers of global elements, and studies stability properties of terminally connected morphisms, including behaviour under bicomma squares.

Significance. If the main theorem is correct, the paper provides a canonical factorization for arbitrary geometric morphisms, unifying the locally connected and essential cases and introducing a useful class of pro-étale morphisms. The accessibility argument in Lemma 3.1.2 and the site-theoretic presentation of cofiltered bilimits are valuable tools. However, the proof of the central factorization is incomplete in a load-bearing way, and several secondary results are sketched rather than fully proved. The contribution is promising and likely correct in broad outline, but the main claim is not established as written.

major comments (3)
  1. [Theorem 3.2.1, proof in §3.2] The proof of terminal connectedness of l_f is incomplete. From a global element b : 1_F → l_f^*(X), the proof constructs a morphism in the oplax colimit of sites, but it never constructs a global element \bar b : 1_B → X in the bilimit topos, nor proves that l_f^*(\bar b) = b, nor verifies uniqueness or naturality in X. The assertion that the codomain object of the constructed morphism is “conveniently sent to 1_F” is not justified: in general l_f^*(E,a,h) = a^*f^*h, which is not the terminal object. Consequently, the natural isomorphism F[1_F, l_f^*(-)] ≅ B[1_B,-] is not established, and terminal connectedness of the left factor remains unproven.
  2. [Proposition 4.2.2, converse direction in §4.2] The converse of the intrinsic characterization is left essentially to the reader. Functoriality of the constructed h^* is explicitly deferred (“We let the reader convince himself”), and lexness is asserted in a few sentences with no details about how preservation of finite limits follows from the colimit decomposition of fibers. Since this proposition underpins the canonical presentation of pro-étale morphisms and is used later, the proof must supply the missing verifications, especially well-definedness with respect to different colimit presentations of the same object.
  3. [Propositions 3.3.3 and 5.3.3, §3.3 and §5.3] The proofs of stability of terminally connected morphisms under bicomma squares rely on a bespoke presentation of the bicomma topos and on the claim that the component at the terminal object can be made an identity in the free lex completion. These steps are not justified in the text. As the stability properties are used to motivate the analogy with the comprehensive factorization system, complete proofs or a clear reference are needed for these results to be accepted.
minor comments (6)
  1. [Section 2.2, opening] Typo: “Alhough” should be “Although”.
  2. [Definition 2.1.10] The phrase “if if it is equivalent” should read “if it is equivalent”.
  3. [Bibliography] The name “Peter Tenant Johnstone” should be corrected to “Peter T. Johnstone” (or the intended spelling). Reference [10] is an informal nLab page; a more standard citation would be preferable.
  4. [§3.2 proof of Theorem 3.2.1] The proof uses the large bilimit bilim_{1_F ↓ f^*} E/E, while Lemma 3.1.2 and Corollary 3.1.3 only establish existence of a small coinitial subcategory. The equivalence between the large and small indexing should be stated explicitly at the point where the bilimit is used.
  5. [Remark 5.2.1] The warning that site-level lifting of global elements is not sufficient in general is welcome, but it highlights a subtlety that the proof of Theorem 3.2.1 does not address: the passage from morphisms in the site to global elements in the sheaf topos requires justification.
  6. [Proposition 5.1.2] The proof states that Pro(E) is “cogenerated from E by cofiltered limits” without a precise statement or reference; a short justification or citation would strengthen the argument.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation found; the central factorization proof is self-contained, with only non-load-bearing self-citations.

full rationale

The main theorem, Theorem 3.2.1, is not obtained by assuming its conclusion. The right factor is constructed as the cofiltered bilimit of etale morphisms indexed by global elements of the inverse image, and the left factor is then shown to be terminally connected by a direct lifting argument on global elements. The existence of the bilimit is justified by Lemma 3.1.2 using local presentability of Grothendieck topoi and the external bilimit theorem of Dubuc-Yuhjtman [4], not by a self-cited uniqueness theorem. The essential case from [2] is presented as motivation and as a special case to which the general factorization restricts, but the proof of Theorem 3.2.1 does not reduce to it. The only self-citations are [2] (Caramello's prior essential factorization and cofinality criterion) and [3] (Di Liberti-Osmond, used in Remark 2.1.15 for the standard fact that Lex has filtered pseudocolimits). Neither is load-bearing for the central argument. The skeptic's objection that the proof of terminal connectedness lacks a fully explicit inverse construction is a possible correctness gap, not a circularity: it concerns an attempted derivation, not a presupposition of the result. Accordingly, no circular step can be exhibited with quotation, and the appropriate score reflects only minor non-load-bearing self-citation.

Assumptions & free parameters 0 free parameters · 5 assumptions · 1 invented entities

No free parameters appear; the paper is a pure mathematical derivation. The central theorem rests on standard topos-theoretic facts and on the cited existence of cofiltered bilimits of topoi. The only new named object is the class of pro-etale morphisms, introduced by definition. No fitted constants or empirical inputs appear.

assumptions (5)
  • standard math Small cofiltered bilimits of Grothendieck topoi exist and are computed as sheaves over filtered pseudocolimits of lex sites.
    Invoked in Section 2.1 (Proposition 2.1.7, from Dubuc and Yuhjtman [4]) and used to build the pro-etale factor.
  • standard math Grothendieck topoi are locally presentable, and every object becomes compact above some cardinal.
    Used in Lemma 3.1.2 to replace the large category of global elements by a small cofinal subcategory.
  • standard math In a bi-orthogonality structure, right classes are closed under bilimits.
    Used in Lemma 3.1.1 to transfer bi-orthogonality from etale morphisms to pro-etale morphisms; relies on [5] and [11].
  • domain assumption The inverse image of a geometric morphism preserves finite limits and colimits.
    Standard definition of geometric morphism used throughout the proof of Theorem 3.2.1.
  • standard math The category of global elements of a flat functor is cofiltered.
    Remark 2.1.3, used to ensure the indexing diagram for the bilimit is cofiltered.
invented entities (1)
  • Pro-etale geometric morphism
    purpose: Right class in the new factorization; a cofiltered bilimit of etale geometric morphisms (Definition 2.1.10).
    Introduced as a definition; the internal theorems justify it, but there is no external falsifiable handle, and its full utility depends on the factorization theorem.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On a (terminally connected, pro-etale) factorization of geometric morphisms." pith.science (2026). https://pith.science/paper/WKBSFO5P

@misc{pith2026250204213,
  author       = {Pith},
  title        = {Pith review of: On a (terminally connected, pro-etale) factorization of geometric morphisms},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WKBSFO5P}},
  note         = {Machine review of arXiv:2502.04213}
}
read the original abstract

We extend the classical (connected, etale) factorization of locally connected geometric morphisms into a (terminally connected, pro-etale) factorization for all geometric morphisms between Grothendieck topoi. We discuss properties of both classes of morphisms, particularly the relation between pro-etale geometric morphisms and the category of global elements of their inverse image; we also discuss their stability properties as well as some fibrational aspects.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

13 extracted references · 11 canonical work pages

  1. [1]

    Handbook of Categorical Algebra: Volume 3, Sheaf Theory

    Francis Borceux. Handbook of Categorical Algebra: Volume 3, Sheaf Theory . Vol. 3. Cam- bridge University Press, 1994

    work page 1994

  2. [2]

    Denseness conditions, morphisms and equivalences of topos es

    Olivia Caramello. Denseness conditions, morphisms and equivalences of topos es. 2020. arXiv: 1906.08737 [math.CT]

    arXiv 2020

  3. [3]

    Bi-accessible and bipresentable 2-categories

    Ivan Di Liberti and Axel Osmond. Bi-accessible and bipresentable 2-categories . 2022. doi: 10.48550/ARXIV.2203.07046. url: https://arxiv.org/abs/2203.07046

    arXiv doi:10.48550/arxiv.2203.07046 2022

  4. [4]

    A construction of 2-cofiltered bilimits of topoi

    Eduardo J. Dubuc and Sergio Yuhjtman. A construction of 2-cofiltered bilimits of topoi . 2011. arXiv: 1107.1685 [math.CT]

    work page Pith review arXiv 2011

  5. [5]

    Proper factorization systems in 2-categories

    Marc Dupont and Enrico Vitale. “Proper factorization systems in 2-categories”. In: Journal of Pure and Applied Algebra 179.1 (2003), pp. 65–86. issn: 0022-4049. doi: https://doi.org/10.1016/S0022-4049( 38

    work page doi:10.1016/s0022-4049( 2003

  6. [6]

    Sketches of an Elephant: A Topos Theory Compendium

    Peter Tenant Johnstone. Sketches of an Elephant: A Topos Theory Compendium . Vol. 1. Oxford logic guides. Oxford University press, 2002

    work page 2002

  7. [7]

    Local maps of to poses

    Peter Tenant Johnstone and Ieke Moerdijk. “Local maps of to poses”. In: Proceedings of the London Mathematical Society s3-58 (2 1989), pp. 281–305. doi: https://londmathsoc.onlinelibrary.wiley.co

    work page 1989

  8. [8]

    Topo-logie

    Andr´ e Joyal and Mathieu Anel. “Topo-logie”. In: New Spaces in Mathematics and Physics . Cambridge University Press, 2019

    work page 2019

Show all 13 references
  1. [9]

    Higher topos theory

    Jacob Lurie. “Higher topos theory”. In: Higher Topos Theory (AM-170). Princeton University Press, 2009

    2009

  2. [10]

    url: https://ncatlab.org/michaelshulman/show/comprehensive+factorization

    Shulman Lab. url: https://ncatlab.org/michaelshulman/show/comprehensive+factorization

  3. [11]

    Variation on a comprehensive theme

    Ross Street. “Variation on a comprehensive theme”. In: arXiv preprint arXiv:2104.02887 (2021)

    2021 arXiv

  4. [12]

    The comprehensive factoriza tion and torsors

    Ross Street and Dominic Verity. “The comprehensive factoriza tion and torsors”. In: Theory Appl. Categ 23 (2010), pp. 42–75

    2010

  5. [13]

    Pro-Categories and Multiadjoint Functors

    Walter Tholen. “Pro-Categories and Multiadjoint Functors”. I n: Canadian Journal of Math- ematics 36.1 (1984), pp. 144–155. doi: 10.4153/CJM-1984-010-2 . Olivia Caramello Dipartimento di Scienza e Alta Tecnologia, Universit `a degli Studi dell’Insubria, via V al- leggio 11, 22...

    1984 doi

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.