On 2-categories of extensions

D. Kaledin

REVIEW 3 major objections 4 minor 8 references

On 2-categories of extensions

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The derived category of an abelian category with enough injectives carries a natural 2-category of extensions, and the enhancement formalism of [Ka4] recovers it exactly.

desk verdict A useful concrete illustration of the enhancement formalism, with a load-bearing final comparison left unproved and an unjustified reduction to the point. read the letter →

arxiv 2505.17286 v1 pith:FBN7SNCO submitted 2025-05-22 math.AG math.CTmath.KT

D. Kaledin This is my paper · ORCID

classification math.AGmath.CTmath.KT MSC 18N1018G8018G35

keywords derivedcategories2-categoriesextensionsenhancementstriangulatedsemicartesianproductsabelianYonedaExt

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

The paper's claim is that the derived category of an abelian category with enough injectives carries a natural 2-category, named the 2-category of extensions, whose objects are complexes of length two and whose morphisms are themselves extensions up to homotopy. The author argues that the triangulated structure alone cannot see this 2-category, because at length two the homotopy type of maps between two objects is a groupoid rather than a set, and a triangulated category only records connected components. The paper first builds this 2-category by hand using explicit chain complexes and splittings, then shows that a short construction in the enhancement formalism of [Ka4] and [Ka5] recovers the same 2-category through a natural 2-equivalence. A sympathetic reader should care because it is a concrete low-complexity case where enhancements provably carry information that triangulated categories forget, and where the enhanced construction is simple enough to see exactly why.

What carries the argument

The load-bearing machinery is the enhanced 2-category $\Delta^h E$ built from any enhanced category $E$ by formula (3.3): take the enhanced relative functor category $\mathrm{Fun}^h(\Delta_q|\Delta,E)$, then cut it down by a semicartesian square using $\epsilon^h_*$. The semicartesian product, an enhanced substitute for a fiber product characterized by a universal property and by being an epivalence (conservative, essentially surjective, and full), is the part the author calls absolutely crucial: for two equal objects and identity homology maps, the natural comparison functor from the hand-built morphism category to the ordinary cartesian product is only an epivalence, with source a groupoid with nontrivial $\pi_1$ and target discrete. The enhanced functor categories supply the homotopy types of mapping objects, and the semicartesian product is what prevents that homotopical information from being collapsed.

What would settle it

For any abelian category $A$ with enough injectives, take $M^q=N^q$ and maps $f_0=f_1=\mathrm{id}$; the paper predicts the morphism category in the enhanced 2-category is a groupoid with fundamental group $\mathrm{Hom}(H_0(M^q),H_1(M^q))$, while the triangulated truncation would be discrete. Computing this groupoid directly from the enhanced functor category, say for modules over a ring with nontrivial $\mathrm{Ext}^1$, and finding it discrete would falsify Proposition 3.1; finding the predicted nontrivial $\pi_1$ confirms the need for enhancement.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is Proposition 3.1: for any abelian category $A$ with enough injectives, the enhanced 2-category $\Delta^h D_{[0,1]}(A)^h$ is 1-truncated and naturally 2-equivalent to the 2-category $C^{(2)}_{[0,1]}(A)$ constructed by hand in Section 2. This means the elementary abelian-category object whose objects are length-2 complexes, whose 1-morphisms are admissible functors from $\Delta[1]$, and whose 2-morphisms are splittings of the resulting extensions is exactly what the enhancement formalism produces. The comparison functor is the quotient $h : K(C_{[0,1]}(A)) \to D_{[0,1]}(A)^h$ applied fiberwise; the proof reduces to showing that over each pair of objects the induced functor on morphism categories is an epivalence, and the nontrivial content is that it is only an epivalence, not an equivalence. The paper also asserts that the triangulated formalism does not deliver this structure, because the needed morphism groupoids have nontrivial $\pi_1$ in general, which a triangulated category cannot see.

Load-bearing premise

The paper rests on the enhancement black boxes imported from [Ka4] and [Ka5]: the existence and universal property of enhanced functor categories, and especially of semicartesian products, which Remark 3.2 calls absolutely crucial; if that machinery fails, the two-line recovery of the 2-category collapses.

Editorial extensions

If this is right

If Proposition 3.1 is correct, the 2-category of extensions is not an ad hoc abelian-category construction: it is forced by the general enhancement formalism, so any enhancement of a derived category that satisfies the [Ka4] axioms will produce it.
The morphism categories of $C^{(2)}_{[0,1]}(A)$ admit an explicit description (Proposition 2.13): the fiber over a pair of maps $f_0,f_1$ is the groupoid $\mathrm{Spl}((f_1\oplus \mathrm{id})\circ(M^q\oplus N^q)\circ(\mathrm{id}\oplus(-f_0)))$, nonempty exactly when the Yoneda $\mathrm{Ext}^2$ classes match, and then it is a gerb over $\mathrm{Ext}^1(H_0(M^q),H_1(N^q))$.
Consequently, whenever $A$ has a derived category the 2-category $C^{(2)}_{[0,1]}(A)$ is bounded, and its truncation is the ordinary full subcategory $D_{[0,1]}(A)$.
For the full derived category $D(A)^h$ rather than $D_{[0,1]}(A)^h$, the same construction still gives an enhanced 2-category, but it is not 1-truncated, and the paper leaves open whether its 1-truncation admits a concise explicit description of the kind given in Section 2.

Reading between the lines

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

Going beyond the paper, one can test the same semicartesian construction on the subcategory of complexes of length at most $n$; the natural expectation is an $n$-truncated higher category refining $D_{[0,n]}(A)$, with the same proof pattern reducing to an epivalence statement.
The appearance of $\mathrm{Ext}^1$ gerbs as morphism categories suggests these 2-categories are a natural home for obstruction-theoretic gluing data, since a morphism itself is an extension and composing such morphisms may encode associativity information that classical triangulated gluing lacks.
If the enhancement technology eventually covers abelian categories without enough injectives, Proposition 3.1 should carry over verbatim; until then, for such categories the enhanced construction is only conjecturally the right refinement.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Kaledin does something concrete here: he builds the 2-category of extensions C^(2)_[0,1](A) by hand, gives a complete description of its morphism categories in Proposition 2.13, and then shows that the enhancement formalism of [Ka4] recovers it. The Section 2 part is self-contained and well executed. The paper is honest that it is an illustration rather than a new general theorem. The point that the triangulated structure alone forgets this 2-category is worth recording, and Remark 3.2's warning that one really needs semicartesian rather than cartesian products is instructive.

The soft spots are concentrated in Proposition 3.1. Two steps are unsecured as written. First, the reduction 'we can replace A with J^o A' is not licensed by the stated hypothesis that A has enough injectives; for arbitrary posets J, J^o A need not have enough injectives, so the enhancement construction of Section 3.1 does not obviously apply. This might be repairable with a weaker hypothesis, but as written it is a gap. Second, the key comparison (3.7) is dismissed as 'straightforward diagram chasing' and left to the reader. That is exactly where the enhancement is shown to reproduce the hand-built morphism categories; essential surjectivity is not established. These are probably fixable, but they make the central equivalence conditional rather than proven.

There is also the dependence on unpublished black boxes from [Ka4] and [Ka5]. That is not a flaw in itself, but it means the paper cannot be checked independently until those references are available.

Who this is for: people working with enhancements and derived categories, especially anyone who wants a concrete example of what the enhancement formalism buys beyond triangulated structure. The Section 2 construction stands on its own and could be cited independently of the enhancement part.

Recommendation: send it to a serious referee. The referee should ask for a complete proof of the (3.7) epivalence and for a fix or clarification of the J^o A reduction. The paper deserves referee time: the core idea is sound and the explicit construction is valuable. My own verdict would be conditional acceptance pending those repairs.

Referee Report

3 major / 4 minor

Summary. The paper introduces a 2-category of extensions C^(2)_[0,1](A) for an abelian category A, whose objects are two-term chain complexes, whose 1-morphisms are complex maps, and whose 2-morphisms are given by splittings of a certain exact sequence (Proposition 2.13). The construction in Section 2 is explicit and uses only abelian-category techniques, via admissible functors and the Segal-category formalism. In Section 3, assuming A has enough injectives, the paper uses the enhancement formalism of [Ka4] to define an enhanced 2-category Δ^h D[0,1](A)^h and claims (Proposition 3.1) that it is 1-truncated and naturally 2-equivalent to C^(2)_[0,1](A). The paper argues that the triangulated structure alone does not deliver this 2-category, and that enhancement is necessary. The comparison in Proposition 3.1 is the central technical claim.

Significance. Section 2 provides a careful and mostly self-contained description of the morphism categories of the extension 2-category; Proposition 2.13 is worked out in detail and yields an explicit formula in terms of splittings and Yoneda Ext. If the comparison in Proposition 3.1 can be fully proved, the paper would give a clean illustration of the enhanced-category formalism, showing that it recovers exactly the higher structure that a triangulated structure forgets, with a concrete example (Remark 3.2) demonstrating why semicartesian products are necessary. The paper is honest about its limitations, explicitly flagging black-box dependencies and the remaining verification. However, as written, the main theorem is conditional on two unproved steps and on nontrivial results from the author's earlier preprints.

major comments (3)

[§3.2 (proof of Prop. 3.1)] The reduction "since we can replace A with J^o A, it suffices to consider the situation over pt" is not justified by the hypothesis that A has enough injectives. For an arbitrary J∈Pos, J^o A need not have enough injectives (e.g., A = torsion abelian groups has enough injectives, but functor categories over infinite posets need not), yet the enhancement D[0,1](−)^h was constructed in §3.1 only under the enough-injectives assumption. The proof must either strengthen the assumption to A Grothendieck or provide a naturality argument that avoids needing D(J^o A) constructed via h-injective complexes.
[§3.2 (Eq. (3.7))] The assertion that the functor (3.7) is an epivalence is the load-bearing step in the proof of Proposition 3.1, but it is dismissed as "straightforward diagram chasing" and left to the reader. In particular, essential surjectivity is not shown: given an arbitrary object in D[0,1](Fun([1],A))_{M,N}, the existence of a splitting of (f1⊕ id)◦(M⊕N)◦(id⊕(−f0)) and the associated zigzag (3.8) is exactly what needs to be proved. Full faithfulness on 2-morphisms is also not demonstrated. Please include a complete proof of this comparison, or state and prove it as a separate lemma.
[§3.1 (facts (i), (ii))] The proof of Proposition 3.1 rests on substantial results imported from the author's preprints [Ka4], [Ka5]: existence of enhanced functor categories and of semicartesian products with the stated universal properties, plus the identification of special functors used in (3.4). The paper does not state these results precisely, so a reader cannot verify the argument without consulting a lengthy preprint. Please state the exact black-box results (or provide an appendix with their statements) and indicate which parts of Proposition 3.1 depend on each.

minor comments (4)

[§2.3 (Def. 2.12)] In the sentence defining the categories of morphisms, "C[0,A]" should be "C[0,1](A)".
[§3.2 (Prop. 3.1)] Proposition 3.1 refers to "Corollary 2.11" for the 2-category C^(2)_[0,1](A), but Corollary 2.11 only establishes the Segal property; the 2-category itself is introduced in Definition 2.12. Please adjust the reference.
[§2.3 (proof of Prop. 2.13)] The notation M∆q, N∆q is used without an explicit definition; spell out that these are the admissible constant functors corresponding to M q and N q under Example 2.7.
[§3.2 (Remark 3.2)] In the example with M q=N q and f0=id, f1=id, it would be helpful to state explicitly that the non-trivial π1 of the source is isomorphic to Hom(H0(M q), H1(N q)), as follows from Proposition 2.13.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: Section 3's self-citations are load-bearing but the target equivalence is not built into the cited black boxes.

full rationale

The paper's Section 2 constructs the 2-category C^(2)_[0,1](A) self-containedly from abelian-category data (admissible functors over twisted arrow categories, splittings, and the Segal condition), and Section 3's Proposition 3.1 is a comparison theorem between this hand-built 2-category and the enhanced-category construction Δ^hD[0,1](A)^h. No equation in the paper defines either side in terms of the other, and no fitted parameter is renamed as a prediction. The proof does rely on black-box results from the author's earlier [Ka4]/[Ka5] (enhanced functor categories, semicartesian products, [Ka4, Lemma 7.3.3.7], [Ka4, Prop 7.5.6.2]), and these citations are load-bearing for the enhanced half; however, they are external prior results rather than assumptions that already contain the target equivalence. The two genuine weaknesses are not circularity: (i) in the proof of Proposition 3.1, the reduction 'since we can replace A with J^o A' is not licensed by the stated 'enough injectives' hypothesis for arbitrary posets J, and (ii) the essential epivalence of (3.7) is left as 'straightforward diagram chasing that we leave to the reader.' These are gaps in proof completeness and correctness, not instances of a derivation reducing to its own input. Thus the circularity score is 0.

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

The paper's own mathematical content adds the explicit 2-category and the comparison, but it pulls a substantial black-box technology from the author's unpublished enhancement program; no numerical parameters are fitted and no new postulates beyond the enhancement framework are introduced.

assumptions (4)

ad hoc to paper The enhanced category formalism of [Ka4] is correct as used: enhanced functor categories and semicartesian products exist with the stated universal properties.
Section 3.1 takes facts (i) and (ii) as black boxes, citing [Ka4, Corollary 7.3.3.5, Lemma 7.3.3.7] and [Ka5]; Remark 3.2 says the semicartesian product is 'absolutely crucial'.
domain assumption A has enough injectives, and for every J in Pos, J^o A also has enough injectives (or the enhancement D(A)^h is defined for diagram categories of the form J^o A).
Section 3.1 says 'for simplicity, assume that A has enough injectives'; Section 3.2 reduces to the pt case by replacing A with J^o A for arbitrary J in Pos.
domain assumption A admits a derived category D(A) and an enhancement D(A)^h realized through h-injective complexes.
Section 3.1 constructs D(A) as a full subcategory of the chain-homotopy category Ho(A) spanned by h-injective complexes, and D[0,1](A) inherits an enhancement.
standard math Standard homological algebra and category-theoretic facts: Dold-Kan equivalence, Grothendieck fibrations, Segal conditions, Kan extensions, and the four-term exact sequence associated to a two-term complex.
Used throughout Sections 1 and 2, for example (2.3), (1.1), and Proposition 2.8.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On 2-categories of extensions." pith.science (2026). https://pith.science/paper/FBN7SNCO

@misc{pith2026250517286,
  author       = {Pith},
  title        = {Pith review of: On 2-categories of extensions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FBN7SNCO}},
  note         = {Machine review of arXiv:2505.17286}
}

read the original abstract

This is essentially an illustration for the general technology of homotopical enhancements developed recently in arxiv:2409.17489. We take the derived category of an abelian category, and we look at the full subcategory spanned by complexes of length 2. This has a natural refinement to a 2-category that we call "the 2-category of extensions". However, just using the triangulated structure on the derived category is not enough to obtain this refinement. In this short note, we first construct the 2-category of extensions by hand -- that is, using abelian category techniques -- and then show how it can be recovered very easily and naturally in the enhanced formalism of arxiv:2409.17489.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

8 extracted references · 7 canonical work pages

[1]

Dwyer, Hirshhorn, D

W. Dwyer, Hirshhorn, D. Kan, and J. Smith, Homotopy limit functors on model categories and homotopical categories , Mathematical Surveys and Monographs, 113 , AMS, Providence, RI, 2004

work page 2004
[2]

Grothendieck, Cat\'egories fibr\'ees et descente , Expos\'e VI in SGAI

A. Grothendieck, Cat\'egories fibr\'ees et descente , Expos\'e VI in SGAI

work page
[3]

Kaledin, How to glue derived categories , Bull

D. Kaledin, How to glue derived categories , Bull. Math. Sci. 8 (2018), 477--602

work page 2018
[4]

Kaledin, Trace theories, B\"okstedt periodicity and Bott periodicity , arxiv:2004.04279

D. Kaledin, Trace theories, B\"okstedt periodicity and Bott periodicity , arxiv:2004.04279

arXiv 2004
[5]

Kaledin, Adjunction in 2 -categories , Russian Math

D. Kaledin, Adjunction in 2 -categories , Russian Math. Surveys 75 (2020), 883-927

work page 2020
[6]

Kaledin, Enhancement for categories and homotopical algebra , arxiv:2409.17489

D. Kaledin, Enhancement for categories and homotopical algebra , arxiv:2409.17489

work page arXiv
[7]

Kaledin, how to enhance categories, and why? , to appear in Russian Math

D. Kaledin, how to enhance categories, and why? , to appear in Russian Math. Surveys

work page
[8]

J.-L. Verdier, Des cat\'egories d\'eriv\'es des cat\'egories ab\'eliennes , with a preface by Luc Illusie, edited and with a note by Georges Maltsiniotis, Ast\'erisque 239 (1963, published in 1996)

work page 1963

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Dwyer, Hirshhorn, D

W. Dwyer, Hirshhorn, D. Kan, and J. Smith, Homotopy limit functors on model categories and homotopical categories , Mathematical Surveys and Monographs, 113 , AMS, Providence, RI, 2004

work page 2004

[2] [2]

Grothendieck, Cat\'egories fibr\'ees et descente , Expos\'e VI in SGAI

A. Grothendieck, Cat\'egories fibr\'ees et descente , Expos\'e VI in SGAI

work page

[3] [3]

Kaledin, How to glue derived categories , Bull

D. Kaledin, How to glue derived categories , Bull. Math. Sci. 8 (2018), 477--602

work page 2018

[4] [4]

Kaledin, Trace theories, B\"okstedt periodicity and Bott periodicity , arxiv:2004.04279

D. Kaledin, Trace theories, B\"okstedt periodicity and Bott periodicity , arxiv:2004.04279

arXiv 2004

[5] [5]

Kaledin, Adjunction in 2 -categories , Russian Math

D. Kaledin, Adjunction in 2 -categories , Russian Math. Surveys 75 (2020), 883-927

work page 2020

[6] [6]

Kaledin, Enhancement for categories and homotopical algebra , arxiv:2409.17489

D. Kaledin, Enhancement for categories and homotopical algebra , arxiv:2409.17489

work page arXiv

[7] [7]

Kaledin, how to enhance categories, and why? , to appear in Russian Math

D. Kaledin, how to enhance categories, and why? , to appear in Russian Math. Surveys

work page

[8] [8]

J.-L. Verdier, Des cat\'egories d\'eriv\'es des cat\'egories ab\'eliennes , with a preface by Luc Illusie, edited and with a note by Georges Maltsiniotis, Ast\'erisque 239 (1963, published in 1996)

work page 1963