Local mollification of metrics with small curvature concentration

Man-Chun Lee; Tang-Kai Lee

REVIEW 3 major objections 4 minor 9 references

Local mollification of metrics with small curvature concentration

T0 review · 3 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read Small curvature concentration alone yields local Ricci smoothing

desk verdict A real advance—local Ricci-flow smoothing without any Ricci lower bound—but the base of the induction is a black-box Sobolev-to-entropy lemma whose hypotheses are never stated, and the referee needs to verify that bridge. read the letter →

arxiv 2510.12673 v2 pith:3UXTJKBE submitted 2025-10-14 math.DG

Man-Chun Lee , Tang-Kai Lee This is my paper · ORCID

classification math.DG MSC 53E2053C2153C23

keywords curvatureconcentrationRicciflowlocalsmoothinglog-SoboleventropySobolevinequalityGromov-HausdorffcompactnesstopologicalgaptheoremAhlforsregularity

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 a metric with small local L^(n/2) curvature concentration can be smoothed by the Ricci flow for a uniform short time, assuming only Euclidean-type local volume growth and a uniform Sobolev inequality on unit balls. No lower bound on Ricci curvature is needed, removing the condition that anchored all prior local smoothing results. The flow has instantaneous curvature bound proportional to ε/t, injectivity radius at least √t, and distances Hölder-equivalent to the original metric. The authors use this to show that Gromov–Hausdorff limits of compact manifolds with bounded Sobolev constant, volume growth, and curvature concentration are topological manifolds outside finitely many points, and that complete manifolds with Euclidean Sobolev inequality, Euclidean volume growth, and small total curvature concentration are diffeomorphic to R^n. They also flag open points: whether the singular set in the compactness limit is always smooth, and whether curvature concentration can jump when the initial metric has unbounded curvature.

What carries the argument

The central object is the local log-Sobolev entropy functional ν(Ω,g,τ), which measures the best weighted log-Sobolev constant of a domain at scale τ. The proof converts the initial unit-ball Sobolev inequalities into a uniform lower bound on this entropy, then runs four intertwined estimates: a persistence lemma showing that small curvature concentration remains small while an entropy lower bound holds; a priori bounds converting volume growth, Sobolev control, and small initial concentration into improved curvature decay; an inductive extension scheme that patches local Ricci flows across time steps using a short-time existence theorem for incomplete metrics with bounded curvature; and a r

What would settle it

Find a sequence of metrics satisfying the local volume growth and unit-ball Sobolev bounds with ε → 0 whose Ricci flow, within the claimed time T̂, has a point where the curvature at time t exceeds Λ1 ε/t while the initial concentration on that ball is below ε. Equivalently, compute the local log-Sobolev entropy on unit balls of such a sequence: if the entropy is not uniformly bounded below as ε → 0, the Sobolev-to-entropy lemma used in the proof fails.

Watch

Extended reading notes

Core claim

The central claim, stated as Theorem 1.1, is that under the conditions Vol_{g0}(B(x,r)) ≤ v0 r^n, a unit-ball Sobolev inequality with constant Cs, and ||Rm(g0)||_{L^(n/2)(B(x,1))} ≤ ε with ε below a threshold δ1, a smooth Ricci flow exists on B_{g0}(x0,1+r̄)×[0,T̂] starting at g0, with |Rm(g(t))| ≤ Λ1 ε/t, injectivity radius at least √t, local entropy bounded below, and distances at positive times Hölder-equivalent to the initial ones, with constants depending only on n, v0, and Cs. Because the hypotheses are purely local and involve no Ricci lower bound, the theorem is a local mollification statement. Global versions produce a complete flow on non-compact manifolds and yield two geometric a

Load-bearing premise

The proof depends on converting unit-ball Sobolev inequalities into uniform lower bounds on the local log-Sobolev entropy, via a quoted lemma; if that Sobolev-to-entropy bridge fails at some scale, or if the entropy lower bound is lost during the inductive patching, the local smoothing does not go through.

Editorial extensions

If this is right

On every ball where the three hypotheses hold, the metric admits a uniform-time Ricci flow whose time-t metric has pointwise curvature at most Λ1 ε/t and injectivity radius at least √t.
A sequence of compact manifolds with uniformly bounded Sobolev constant, at-most-Euclidean volume growth, bounded L^(n/2) curvature norm, and bounded diameter subsequentially converges in the Gromov–Hausdorff sense to a space that is a topological manifold away from finitely many singular points.
If the global L^(n/2) curvature norm of the sequence is below the threshold in Theorem 1.1, the singular set is empty and the limit carries a smooth structure.
A complete non-compact manifold with a Euclidean Sobolev inequality, Euclidean volume growth, and sufficiently small total L^(n/2) curvature concentration has a global Ricci flow with curvature decay ε/t and injectivity radius at least √t, and is diffeomorphic to R^n.
The long-time version of the flow keeps the local curvature concentration bounded by ε along the flow, and the curvature eventually decays faster than t^(-1), which is the input that forces the Euclidean topology.

Reading between the lines

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

A natural test is whether the local smoothing remains stable under Gromov–Hausdorff limits; if so, the finite singular set in the compactness theorem should be removable, upgrading the topological manifold structure to a smooth structure, as the authors conjecture.
The same Sobolev-to-entropy bridge could be applied to other scale-invariant integral curvature controls, such as norms of the trace-free Ricci or Weyl curvature, yielding analogous local smoothing under curvature concentration alone.
In odd dimensions the paper's arguments do not settle whether the smallness assumption on ||Rm||_{L^(n/2)} is necessary for the diffeomorphic conclusion; an odd-dimensional example with small but nonzero total curvature and Euclidean volume growth would clarify the gap phenomenon.
If the volume-growth hypothesis in the complete non-compact theorem can be replaced by a Kato-class decay of the negative part of Ricci, as the authors suggest, the conclusion would become a purely Sobolev-plus-curvature rigidity statement.

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.

Desk Editor's Note

read the letter

Colleague,

You should know upfront: this paper does what it says. The main theorem—local smoothing under unit-ball Sobolev bounds, volume growth, and small L^{n/2} curvature concentration, with no Ricci lower bound—is a genuine extension of [CHL24], and the compactness and Euclidean-diffeomorphism applications are new. The induction in Section 3 is carefully assembled: each extension step is designed to preserve curvature decay, entropy lower bound, injectivity radius, and the radius decay, and the constants are explicit enough to see the logic.

The soft spot is exactly the one flagged in the stress-test. The base case of the induction is the conversion of a unit-ball Sobolev inequality into a lower bound on the local ν-entropy, done by invoking [Mar25a, Lemma 3.2]. The manuscript gives neither the statement nor the hypotheses of that lemma. It is not a small technicality: the entropy bound is the starting point for Lemma 2.1, Proposition 2.4, and the inductive control (III). If Lemma 3.2 requires a Ricci lower bound, an injectivity radius, or a stronger smallness condition than the ε allowed here, the whole construction has no base. I read the passage carefully; the paper simply says 'By [Mar25a, Lemma 3.2]' with no further detail. The onus is on the authors to state the lemma and verify its assumptions are satisfied. This is the main reason I would not call the paper fully verified.

There are smaller issues. Theorem 5.1 states constants independent of v0, but the proof uses v0; either the constants should include v0 or the volume growth assumption should be removed or absorbed. That's a fixable omission, not a fatal flaw. The final diffeomorphism conclusion also leans on [HP25], another recent external result, so the gap theorem inherits that dependency. The citation pattern is otherwise legitimate: self-citations are used as analytic tools, not as the target conclusion.

Do not desk reject this. It is exactly the kind of paper that needs a referee who can check the external lemmas. Send it out, and make sure the referee has access to [Mar25a]. If the bridge holds, this is a strong result.

Referee Report

3 major / 4 minor

Summary. The paper proves a local Ricci-flow smoothing theorem (Theorem 1.1) for manifolds with small L^{n/2} curvature concentration, assuming only a local Sobolev inequality and an upper volume-growth bound, with no Ricci curvature lower bound. The proof constructs a local Ricci flow by an induction that alternately improves curvature estimates and extends the flow, using local entropy as the main mechanism. The paper then applies this smoothing to prove a compactness theorem for manifolds with bounded curvature concentration (Theorem 1.3) and a gap theorem showing that complete manifolds with Euclidean Sobolev inequality, Euclidean volume growth, and small total curvature concentration are diffeomorphic to R^n (Theorem 1.4).

Significance. If the central theorem is correct, it is a substantial advance: prior Ricci-flow smoothing results in this direction required a Ricci lower bound, and the present paper removes it while also localizing the argument. The proof is nontrivial and carefully structured, with an induction that explicitly tracks constants, and the paper is honest about several open questions. The applications to compactness and rigidity are natural and would be influential. The main caveats are that several load-bearing analytic inputs are invoked from very recent external sources without statements, and one estimate in Theorem 1.1 appears not to follow from the written proof as quantified. These issues are fixable in principle, but they need to be addressed before the central claims can be accepted.

major comments (3)

[Theorem 1.1(1), §3 (constants list and Claim 3.3)] Theorem 1.1 asserts |Rm(g(t))| ≤ Λ1 ε t^{-1} for every ε∈(0,δ1], with Λ1 independent of ε. In the proof, however, the constant ε1 is chosen before ε and is independent of it, and Claim 3.3 concludes |Rm(z,t)| ≤ 2ε1Λ0 t^{-1}. Since ε can be arbitrarily small compared with ε1, no fixed Λ1 can make 2ε1Λ0 ≤ Λ1 ε uniformly. The final 'relabelling' cannot repair this, because the theorem's constants are not allowed to depend on ε. The proof as written would establish the weaker bound |Rm(g(t))| ≤ Λ t^{-1}; if (1) is intended, the proof must be reworked to produce a genuinely ε-proportional estimate, or the statement must be changed. The same issue affects Theorem 3.11(1).
[Proof of Theorem 1.1, first paragraph; Lemmas 2.1, 2.16] The proof opens by invoking [Mar25a, Lemma 3.2] to obtain ν(B_{g0}(x,1),g0,2) ≥ −A0(n,Cs) from the local Sobolev inequality. This lemma is not stated, and its hypotheses are not verified. This entropy lower bound is the base case of the induction: it is propagated by Lemma 2.16 and used in Lemma 2.1, Proposition 2.4, and Proposition 2.15. If [Mar25a, Lemma 3.2] requires any extra hypothesis beyond (a)–(c)—for example a lower volume bound, an injectivity or pointwise curvature control, or a stronger smallness condition on ||Rm||_{L^{n/2}}—then the induction has no base case. Similarly, Lemma 2.1 relies on [Mar25a, Lemma 3.3(2)] to convert the entropy bound into a Sobolev inequality at later times; that lemma is also not stated. Please state both lemmas and explicitly verify their hypotheses under exactly (a)–(c).
[Theorem 1.4, §5] The proof of Theorem 1.4 is a one-line appeal to [HP25, Theorem 1.1] after Theorem 5.1. The diffeomorphism-to-R^n conclusion is therefore entirely outsourced to an external result whose precise hypotheses are not given. If [HP25, Theorem 1.1] requires conditions not met by the flow constructed in Theorem 5.1—for instance a specific global curvature decay or a control on the full curvature tensor beyond |Rm(g(t))| ≤ Λε t^{-1}—then Theorem 1.4 does not follow. State the quoted theorem and verify its hypotheses explicitly.

minor comments (4)

[Throughout] The symbol δ1 is used both for the threshold in Theorem 1.1 and for a constant in Lemma 2.1 with a different role; this is confusing and should be relabelled.
[§3, constants list] The notation ε1 is introduced as a threshold, but later in Claim 3.3 it appears in the curvature estimate in a way that suggests it might be the input ε. Please clarify the intended roles of ε and ε1.
[§2, Lemma 2.1] The statement of Lemma 2.1 uses δ1 for the smallness constant but does not explicitly connect it to the δ1 later used in Proposition 2.4. Adding a sentence such as 'where δ1(n,A) is the constant from Lemma 2.1' in Proposition 2.4 would help, as would stating the dependence of δ1 in Lemma 2.1.
[§3, Claim 2.6] In the proof of Claim 2.6, the line 'if α0 ≤ min{(8C0)^{-2} T̃ δ1, C0^{-2}(4C1/3)^{-2} δ^2}' appears to contain a typo: the final δ^2 should likely be δ1^2.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the smoothing theorem is proved by induction over external analytic lemmas, not by reducing its conclusion to its own assumptions.

full rationale

The derivation chain in Theorem 1.1 converts hypotheses (a)-(c) into an initial entropy lower bound through the external lemma [Mar25a, Lemma 3.2], then propagates that entropy bound by induction (statements P(k), Claims 3.3-3.4) using a priori estimates (Lemma 2.1, Propositions 2.4 and 2.15, Lemma 2.16). No step fits a parameter to the target quantity and renames it a prediction, and no equation defining the output is the same equation used as input. The curvature decay |Rm|≤α0 t^{-1} and entropy lower bound (III) are proved by induction from the initial bounded-curvature flow supplied by [LT22, Prop. 4.2], not assumed. The initial entropy bound is imported from [Mar25a], an external source, and the reverse Sobolev-from-entropy implication used inside the induction is likewise external. Self-citations are present ([CHL24], [CL24], [LT22], [Lee25], [CCL22]), but they are cited for specific analytic constructions and uniqueness facts, not for the theorem being proved; in particular [Lee25] is used only to identify two already-constructed local flows on overlaps in Theorem 5.1. There are unverified external hypotheses (e.g., Remark 2.3 acknowledges that the δ1-smallness condition in [Mar25a, Lemma 3.3] is assumed implicitly; the applicability of [Mar25a, Lemma 3.2] to the full hypotheses of Theorem 1.1 is not stated in the paper). These are correctness risks, not circularity: nothing in the paper shows by construction that a conclusion is its own premise. The Holder distance estimate is imported from [Mar25b]/[CHL24], also without circular reduction. Overall this is a standard analytic argument modulo external lemmas, so the circularity score is 0.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

No new particles, forces, functions, or geometric objects are introduced beyond standard Ricci flow, distance balls, and Perelman entropy. The proof's constants are existential outputs of the estimates, not free parameters fitted to data.

assumptions (6)

standard math Perelman's local ν-entropy monotonicity along bounded-curvature Ricci flow and its localized refinements ([Per02], [Wan18], [Che25, Lemma 2.5]).
Used to propagate entropy lower bounds from t=0 to later times, especially in Lemma 2.16, (3.5), and (3.7); without it the inductive entropy control (III) in Theorem 1.1 fails.
domain assumption Conversion between local Sobolev constants and log-Sobolev entropy with explicit constants ([Mar25a, Lemmas 3.2 and 3.3], based on [Ye15]).
Assumption (b) is converted into the entropy lower bound ν≥−A0 that the local Ricci flow construction requires; this is a black-box bridge to the existing literature.
domain assumption Hochard-style short-time existence of complete Ricci flow from bounded pointwise curvature on possibly incomplete domains ([Hoc19], [LT22, Proposition 4.2]).
Used in Claim 3.4 to extend the flow onto shrinking balls; supplies the complete solutions to which entropy estimates are applied.
standard math Ball inclusion and distance-distortion estimates for Ricci flows under upper Ricci bounds ([ST22, Lemma 4.1 and Corollary 3.3]).
Controls how balls shrink under the flow, needed for the induction on radii and for the Hölder distance equivalence in Theorem 1.1.
domain assumption Uniqueness of Ricci flow with scaling-invariant estimates ([Lee25]).
Identifies the rescaled local flows in the proof of Theorem 5.1 and extends them to a single long-time solution.
domain assumption Diffeomorphism criterion from long-time Ricci flow with o(1)/t curvature decay and injectivity lower bound ([HP25, Theorem 1.1]).
This is the final step translating Theorem 5.1 into the topological conclusion of Theorem 1.4; it is cited rather than proved in this paper.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Local mollification of metrics with small curvature concentration." pith.science (2026). https://pith.science/paper/3UXTJKBE

@misc{pith2026251012673,
  author       = {Pith},
  title        = {Pith review of: Local mollification of metrics with small curvature concentration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3UXTJKBE}},
  note         = {Machine review of arXiv:2510.12673}
}

abstract

In this work, we establish a local smoothing result on metrics with small curvature concentration with respect to Sobolev constants and volume growth. In contrast with all previous works, we remove the Ricci curvature condition and completely localize the smoothing. As an application, we prove the compactness of the space of compact manifolds with bounded curvature concentration under Ahlfors $n$-regularity and bounded Sobolev constant. In the complete non-compact case, we show that manifolds with Euclidean type Sobolev inequality, Euclidean volume growth, and small curvature concentration are necessarily diffeomorphic to Euclidean spaces.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

9 extracted references · 2 linked inside Pith

[1]

[And89] M. T. Anderson,Ricci curvature bounds and Einstein metrics on compact manifolds, J. Amer. Math. Soc. 2(1989), no. 3, 455–490. [And90] ,Convergence and rigidity of manifolds under Ricci curvature bounds, Invent. Math.102(1990), no. 2, 429–445. [AC91] M. T. Anderson and Jeff Cheeger,Diffeomorphism finiteness for manifolds with Ricci curvature and Ln...

1989
[10]

Chau and A

[CM24] A. Chau and A. Martens,Long-time Ricci flow existence and topological rigidity from manifolds with pinched scale-invariant integral curvature(2024), available atarXiv:2403.02564. [CTY11] A. Chau, L.-F. Tam, and C. Yu,Pseudolocality for the Ricci flow and applications, Canad. J. Math.63 (2011), no. 1, 55–85. [Che70] J. Cheeger,Finiteness theorems fo...

arXiv 2024
[17]

Ledoux,On manifolds with non-negative Ricci curvature and Sobolev inequalities, Comm

[Led99] M. Ledoux,On manifolds with non-negative Ricci curvature and Sobolev inequalities, Comm. Anal. Geom. 7(1999), no. 2, 347–353. [Lee25] M.-C. Lee,Uniqueness of Ricci flow with scaling invariant estimates(2025), available atarXiv:2503. 20292. [LNN23] M.-C. Lee, A. Naber, and R. Neumayer,d p-convergence andϵ-regularity theorems for entropy and scalar ...

arXiv 1999
[29]

[CHL24] P.-Y. Chan, S. Huang, and M.-C. Lee,Manifolds with small curvature concentration, Ann. PDE10(2024), no. 2, Paper No. 23,

2024
[30]

[Mar25b] ,Removing scalar curvature assumption for Ricci flow smoothing, Bull. Lond. Math. Soc.57(2025), no. 7, 1968–1989. [MS73] J. H. Michael and L. M. Simon,Sobolev and mean-value inequalities on generalized submanifolds ofR n, Comm. Pure Appl. Math.26(1973), 361–379. [Nak88] H. Nakajima,Hausdorff convergence of Einstein 4-manifolds, J. Fac. Sci. Univ....

arXiv 2025
[31]

Chan and M.-C

[CL24] P.-Y. Chan and M.-C. Lee,Gap theorem on manifolds with small curvature concentration, Math. Z.308 (2024), no. 1, Paper No. 9,

2024
[1999]

Based on the 1981 French original; with appendices by M. Katz, P. Pansu, and S. Semmes; translated by Sean Michael Bates. [HL21] F. He and M.-C. Lee,Weakly PIC1 manifolds with maximal volume growth, J. Geom. Anal.31(2021), no. 11, 10868–10885. [Hoc19] R. Hochard,Th´ eor` emes d’existence en temps court du flot de Ricci pour des vari´ et´ es non-compl` ete...

1981
[2008]

[Col97] T. H. Colding,Ricci curvature and volume convergence, Ann. Math.145(1997), 477–501. [DW] X. Dai and G. Wei,Comparison Geometry for Ricci Curvature. Book available at web.math.ucsb.edu/ dai/Ricci-book.pdf. [Gal88] S. Gallot,Isoperimetric inequalities based on integral norms of Ricci curvature, Ast´ erisque157-158(1988), 191–216. Colloque Paul L´ ev...

1997

Show all 9 references

[2019]

Huang and Z

[HP25] S. Huang and Z. Peng,A note on a diffeomorphism criterion via long-time Ricci flow(2025), available at arXiv:2509.05802. [JW25] W. Jiang and G. Wei,Finite diffeomorphism theorem for manifolds with lower Ricci curvature and bounded energy, Math. Z.309(2025), no. 1, Paper No. 16,

2025 arXiv

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [1]

[And89] M. T. Anderson,Ricci curvature bounds and Einstein metrics on compact manifolds, J. Amer. Math. Soc. 2(1989), no. 3, 455–490. [And90] ,Convergence and rigidity of manifolds under Ricci curvature bounds, Invent. Math.102(1990), no. 2, 429–445. [AC91] M. T. Anderson and Jeff Cheeger,Diffeomorphism finiteness for manifolds with Ricci curvature and Ln...

1989

[2] [10]

Chau and A

[CM24] A. Chau and A. Martens,Long-time Ricci flow existence and topological rigidity from manifolds with pinched scale-invariant integral curvature(2024), available atarXiv:2403.02564. [CTY11] A. Chau, L.-F. Tam, and C. Yu,Pseudolocality for the Ricci flow and applications, Canad. J. Math.63 (2011), no. 1, 55–85. [Che70] J. Cheeger,Finiteness theorems fo...

arXiv 2024

[3] [17]

Ledoux,On manifolds with non-negative Ricci curvature and Sobolev inequalities, Comm

[Led99] M. Ledoux,On manifolds with non-negative Ricci curvature and Sobolev inequalities, Comm. Anal. Geom. 7(1999), no. 2, 347–353. [Lee25] M.-C. Lee,Uniqueness of Ricci flow with scaling invariant estimates(2025), available atarXiv:2503. 20292. [LNN23] M.-C. Lee, A. Naber, and R. Neumayer,d p-convergence andϵ-regularity theorems for entropy and scalar ...

arXiv 1999

[4] [29]

[CHL24] P.-Y. Chan, S. Huang, and M.-C. Lee,Manifolds with small curvature concentration, Ann. PDE10(2024), no. 2, Paper No. 23,

2024

[5] [30]

[Mar25b] ,Removing scalar curvature assumption for Ricci flow smoothing, Bull. Lond. Math. Soc.57(2025), no. 7, 1968–1989. [MS73] J. H. Michael and L. M. Simon,Sobolev and mean-value inequalities on generalized submanifolds ofR n, Comm. Pure Appl. Math.26(1973), 361–379. [Nak88] H. Nakajima,Hausdorff convergence of Einstein 4-manifolds, J. Fac. Sci. Univ....

arXiv 2025

[6] [31]

Chan and M.-C

[CL24] P.-Y. Chan and M.-C. Lee,Gap theorem on manifolds with small curvature concentration, Math. Z.308 (2024), no. 1, Paper No. 9,

2024

[7] [1999]

Based on the 1981 French original; with appendices by M. Katz, P. Pansu, and S. Semmes; translated by Sean Michael Bates. [HL21] F. He and M.-C. Lee,Weakly PIC1 manifolds with maximal volume growth, J. Geom. Anal.31(2021), no. 11, 10868–10885. [Hoc19] R. Hochard,Th´ eor` emes d’existence en temps court du flot de Ricci pour des vari´ et´ es non-compl` ete...

1981

[8] [2008]

[Col97] T. H. Colding,Ricci curvature and volume convergence, Ann. Math.145(1997), 477–501. [DW] X. Dai and G. Wei,Comparison Geometry for Ricci Curvature. Book available at web.math.ucsb.edu/ dai/Ricci-book.pdf. [Gal88] S. Gallot,Isoperimetric inequalities based on integral norms of Ricci curvature, Ast´ erisque157-158(1988), 191–216. Colloque Paul L´ ev...

1997