REVIEW 3 major objections 9 minor
Two covers suffice: finite étale and radicial checks give full descent
Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →
T0 review · glm-5.2
2026-07-09 19:11 UTC pith:BMTF7NOA
load-bearing objection Clean descendability criterion with real payoff; arguments in §4 are compressed but hold up the 3 major comments →
Descendability and descent in topological weaves
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The central mechanism is a reduction theorem: descendability of all finitely presented surjections reduces to descendability of just finite étale and finite radicial surjections. The proof uses a stratification result from EGA to locally trivialize any finitely presented surjection by these two types of covers, then uses the localization triangle in the weave to glue descendability across strata. Because descendability (unlike mere descent) is preserved by arbitrary exact functors and symmetric monoidal functors, this reduction transfers automatically across different categories of motivic sheaves — once the finite étale and radicial cases are verified, all finitely presented surjections are
What carries the argument
The argument centers on the notion of descendability (due to Mathew): a morphism f: Y→X is descendable in a weave D if the unit object 1_X can be built from f_*(1_Y) using cofibres, fibres, summands, and tensoring. The key technical ingredients are: (1) Lemma 1.23, which shows descendability glues across a closed-open decomposition via the localization triangle, with indices adding; (2) Corollary 1.10, which shows descendability composes under base change; (3) for finite étale covers, the Euler characteristic of the dualizable object f_*(1_Y) must be invertible, which holds in oriented Q-linear weaves; (4) for finite radicial covers, topological invariance (nil-invariance plus conservativity
Load-bearing premise
The proof of the main reduction theorem relies on a classical stratification result from EGA that decomposes any finitely presented surjection of schemes into pieces trivialized by finite radicial then finite étale covers. The gluing step then depends on the localization triangle interacting correctly with the thick subcategories generated by descendable objects in the weave. If this interaction fails for a particular weave, the bootstrapping from local to global descendab
What would settle it
The main theorem would fail if there exists a weave satisfying the localization property where finite étale and finite radicial surjections are descendable but some finitely presented surjection is not — this would mean the gluing step (Lemma 1.23) does not propagate descendability correctly across strata for that weave.
If this is right
- Any Q-linear oriented topological weave automatically satisfies h-descent on algebraic spaces, and v-descent if it satisfies continuity — no further descent verification is needed.
- Rational motivic sheaves SH_{Q,+} satisfy v-descent on all algebraic spaces, recovering and generalizing h-descent results previously known only for quasi-excellent schemes.
- Étale motivic spectra SH_ét satisfy h-descent on noetherian finite-dimensional schemes with uniformly bounded étale cohomological dimension of residue fields, without inverting 2.
- The forgetting supports isomorphism f_! ≃ f_* holds for proper DM morphisms of Artin stacks in SH_{Q,+}, because descendability ensures f_! commutes past the totalization appearing in Čech descent.
Where Pith is reading between the lines
- The reduction to finite étale and finite radicial covers suggests a strategy for proving descent in any new six-functor formalism: verify just these two cases and the localization property, rather than tackling all finitely presented surjections directly.
- The uniform bound on descendability index (Remark 1.24) — depending only on the Krull dimension of the base — could potentially be used to control convergence rates or truncation levels in computational applications of descent in motivic homotopy theory.
- The gap patched in Theorem D (f_! not commuting past totalizations) may recur in other contexts where one attempts to prove isomorphisms by reduction along a cover; descendability provides a systematic way to verify the necessary commutation.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proves a general criterion (Theorem C / Theorem 1.21) for descendability of finitely presented surjections of algebraic spaces in the framework of topological weaves (introduced by the author in [Kha]). The criterion reduces descendability of all finitely presented surjections to that of finite étale and finite radicial surjections, using a stratification argument (EGA IV4, Prop. 17.16.4) and a localization-triangle bootstrapping lemma (Lemma 1.23). The main applications are: (A) rational motivic sheaves SH_{Q,+} satisfy v-descent on all algebraic spaces; (B) étale motivic spectra SH_ét satisfy h-descent on noetherian finite-dimensional schemes with uniformly bounded étale cohomological dimension (and the 2-inverted version under a weaker virtual bound); (D) the forgetting-supports isomorphism f_! ≃ f_* for proper DM morphisms of Artin stacks in SH_{Q,+}. The paper also provides a corrected proof of Theorem D, patching a gap in a prior argument of the author.
Significance. The results are substantial. Theorems A and B recover and extend earlier h-descent results of Cisinski–Déglise and Ayoub for motivic and étale motivic cohomology, removing quasi-excellence hypotheses (Theorem A) and the invertibility-of-2 hypothesis (Theorem B, via the splitting of SH_ét[1/2] in Lemma 4.3 and Bachmann's rigidity). The descendability approach is stronger than mere descent: it yields commutation of totalizations with arbitrary exact functors (Remark 2.11), which is directly exploited in the proof of Theorem D to fix a gap in [Kha3, Thm. A.7] where f_! was not known to commute past Čech totalizations. The framework is abstract (axiomatized via weaves), so the results transfer to any Q-linear oriented topological weave (Corollary 3.4). The proof of Proposition 4.6 involves a nontrivial Postnikov-filtration and Mittag-Leffler argument for p-completed étale sheaves (§4.3, Lemma 4.8), which is a genuine technical contribution.
major comments (3)
- §4.3, proof of (4.6.2), Lemma 4.8: The cohomological dimension bound for Σ^∞_+ U in C = ̂Shv(X_ét)^∧_p is the key technical input for the descendability index bound in Proposition 4.6. The Milnor sequence argument showing H^{c+1}(Σ^∞_+ U; G) ≃ 0 via the Mittag-Leffler condition on {H^c_ét(U; G/p^k G)}_k is correct in spirit, but the transition from derived p-completeness of G to the vanishing of lim^1 requires that the pro-system {H^c_ét(U; G/p^k G)}_k has surjective transition maps. The surjectivity is established using the long exact sequence from 0 → p^k G/p^{k+1} G ↪ G/p^{k+1} G ↠ G/p^k G → 0 and the vanishing of H^{c+1}_ét(U; p^k G/p^{k+1} G). This is fine when G/p^k G is a sheaf of Z/p^k-modules, but the argument should state more explicitly that H^{c+1}_ét(U; p^k G/p^{k+1} G) ≃ 0 follows from the p-torsion property together with the uniform p-étale cohomological dimension bound ≤c
- §5.3, Theorem 5.4: The proof reduces to the case X = [U/G] with G finite linearly reductive, then uses the exact sequence 1 → G_0 → G ↠ H → 1 from [AHR, Thm. 19.9] with G_0 finite radicial and H tame finite étale. The descendability of q: U ↠ [U/G] is then deduced from assumptions (D1) and (D2). However, the factorization q: U ↠ [U/G_0] → [U/G] involves two steps, and the descendability of the composite requires both factors to be descendable. For the second factor [U/G_0] → [U/G] (an H-torsor), (D1) is invoked. For the first factor U ↠ [U/G_0] (finite radicial), (D2) is invoked. The text should clarify that (D2) applies to U ↠ [U/G_0] specifically: the condition requires every prime invertible either in O_M or in End_D(M)(1_M), and the argument that this holds for G_0 of p-power order in characteristic p is given, but the case where M has mixed characteristics (residue fields of varying
- §1.5, proof of Theorem 1.21, algebraic space case: The extension from schemes to algebraic spaces uses a Nisnevich covering p: X' ↠ X with X' a qcqs scheme and a monomorphic splitting sequence ∅ = V_0 ⊆ V_1 ⊆ ⋯ ⊆ V_m = X (citing [HR, Ex. 3.2, Prop. 3.3]). The recursive application of Lemma 1.23 requires that each Z_j = (V_j ∖ V_{j-1})_red is a scheme and that f ×_X Z_j is descendable. The text states this follows from the scheme case. This is correct, but the reader would benefit from a brief clarification that the descendability of f ×_X Z_j (a finitely presented surjection of schemes) is established by the scheme case of the same theorem, which has already been proved — i.e., there is no circularity because the scheme case uses only the stratification from EGA, not the algebraic space case.
minor comments (9)
- Introduction, paragraph on Theorem D: the phrase 'with the following proof (we consider the DM variant)' is slightly awkward; consider rephrasing to clarify that the DM variant is the one being discussed.
- §1.2, Corollary 1.16: the statement references 'B = Spec(Z)' in the proof but B is not introduced in the corollary statement; it appears as a parameter in the hypothesis ('over a noetherian algebraic space S'). The notation should be made consistent.
- §2.1, Lemma 2.1: the functor F_* in (2.2) is defined as a right adjoint, but the notation F^* is also used; the roles of F_*, F^*, F_!, F^! across Lemmas 2.1 and 2.3 should be checked for consistency with the notation in [Kha2].
- §4.1, proof of Theorem 4.1: the reference to [Bac2, Cor. 2.13] for uniformly bounded p-étale cohomological dimension should specify that this applies to odd p in S_0 and all p in S_1, as the text states but the citation could be more precise.
- §4.3, proof of (4.6.2.a): the notation |sk_i(Y_●)| for the geometric realization of the i-skeleton is introduced without a formal definition; a brief pointer to [Lur, Thm. 1.2.4.1] (already cited) would help.
- §5.1, Theorem 5.1: the dévissage theorem [Ry, Thm. D] is applied to the subcategory C_0 ⊆ C; the closure conditions are listed but the third condition (étale neighbourhood) could state more explicitly which closed substack Z is being used (it is Z = (Y' ∖ U)_red lifted to V).
- References: [MT] (Mattis–Tubach) is cited for the vanishing of the η-periodization in SH_ét[1/2]^-; this is a very recent preprint (arXiv:2511.09476). The paper should confirm that this result is not conditional on unpublished conjectures.
- Typo in §3, line after Proposition 3.1: 'sinced is a power of p' should be 'since d is a power of p'.
- §4.3, Lemma 4.8: the notation G//p^k for the cofibre of p^k: G → G is nonstandard; a brief definition (which is provided in the text) would be better placed before first use.
Circularity Check
No significant circularity found; the derivation chain is self-contained against external benchmarks
full rationale
The paper's central result (Theorem 1.21 / Theorem C) reduces descendability of all finitely presented surjections to descendability of finite étale and finite radicial surjections. The proof uses standard external results: [EGA, IV4, Prop. 17.16.4] for stratification, [HR] for Nisnevich covers of algebraic spaces, and [Ry2] for finite presentation approximation. The bootstrapping mechanism (Lemma 1.23, Corollary 1.10, Lemma 1.8) is proved within the paper from the localization triangle and projection formula, not imported by citation. The applications to SH_{Q,+} (§3) and SH_ét (§4) derive finite étale descendability from the Euler characteristic argument (Corollary 1.17, using orientation + Q-linearity, with the key input Lemma 1.12 from [EK]) and finite radicial descendability from topological invariance (Theorem 1.20, citing [EK, Thm. 2.1.1]). The self-citations to [Kha] for the weave framework and [Kha2] for lisse extensions define the setting but do not assume the target results. Theorem D (forgetting supports) explicitly patches a gap in a prior proof [Kha3, Thm. A.7] by using descendability to ensure f_! commutes past totalizations — this is a correction, not a circular restatement. The arguments in §4.2–4.3 (Postnikov filtration, Mittag-Leffler for p-completed sheaves, Bachmann's rigidity) are compressed but logically independent of the paper's own theorems. No step reduces to its inputs by construction, no fitted parameter is renamed as a prediction, and no uniqueness theorem is self-cited to forbid alternatives. The single point of mild concern is that the entire framework of 'topological weaves' is defined in [Kha] by the same author, but this is a definitional framework (like defining a category before proving theorems about it), not a circular argument. Score 1 reflects this minor self-citation for framework setup, which is not load-bearing for the mathematical content of the proofs themselves.
Axiom & Free-Parameter Ledger
axioms (5)
- domain assumption Topological weaves exist as symmetric monoidal presentable stable ∞-categories satisfying the six-functor formalism axioms of [Kha].
- standard math For a finitely presented surjection of qcqs schemes, there exists a stratification by locally closed subschemes admitting finite radicial and finite étale covers after which f has a section ([EGA, IV4, Prop. 17.16.4]).
- domain assumption Bachmann's rigidity equivalence: SH_ét(X)^∧_p ≃ Shv(X_ét)^∧_p when p is invertible on X ([Bac, Thm. 3.1]).
- domain assumption The minus part of SH_ét[1/2] vanishes on noetherian finite-dimensional schemes ([MT, Thm. A]).
- domain assumption Conservativity of stalks for SH_ét on S_0 ([Bac2, Cor. 5.12]).
invented entities (1)
-
Topological weaves
independent evidence
read the original abstract
We prove a criterion for a finitely presented surjection of algebraic spaces to be descendable in a topological weave. We apply this to show that \'etale motivic spectra satisfy $h$-descent on noetherian finite-dimensional schemes with residue fields of uniformly bounded \'etale cohomological dimension. We also show that rational motivic cohomology satisfies arc-descent in weights $\le 1$, and we construct the ``forgetting supports'' isomorphism $f_! \simeq f_*$ for a proper DM-type morphism of Artin stacks, in rational motivic sheaves.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.