REVIEW 2 major objections 5 minor 1 cited by
An Intersection Principle for Mean Curvature Flow
T0 review · 2 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash
Pith's one-line read Two mean curvature flows that meet have an intersection whose Hausdorff dimension never increases, and for level set flows with finitely many singularities this intersection principle exactly characterizes non-fattening.
desk verdict Strong new intersection-principle results for MCF; the proof of Theorem 1.3 drops a multiplicity assumption with a sketch that needs reworking, but the core is credible and worth refereeing. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The machinery is the reduction of intersections to nodal sets of a single linear parabolic function. Whenever two flows are locally graphs $u$ and $v$ over a common plane, their difference $w=v-u$ satisfies a linear parabolic equation $\partial_t w = \partial_i(a^{ij}\partial_j w) + b^j\partial_j w$ whose coefficients inherit uniform ellipticity and Lipschitz regularity from the standard interior estimates for graphical mean curvature flow; the intersection of the flows is exactly the zero set $\{w=0\}$. A nodal-set estimate for such equations (Theorem 2.1, quoted from the literature) bounds $H^{n-1}$ of that zero set. One-sidedness lemmas and the parabolic strong maximum principle convert a zero-measure intersection into strict separation, and a backwards-uniqueness lemma rules out distinct flows merging. For weak solutions, the paper introduces the concept of a localizable level set flow: cutting the flow along a smooth hypersurface whose intersection with the flow has dimension below $n-1$ splits the flow into two independent level set flows. A topological-change theorem for level set flows is then used to show that flows with finitely many singularities are localizable, which lets the nodal-set argument pass through singular times.
What would settle it
Concrete check: construct two properly embedded smooth MCFs, one compact, satisfying Theorem 1.1 with $H^{n-1}(M_{t_0}\cap N_{t_0})=0$ at some $t_0$ but $M_t\cap N_t\neq\emptyset$ for some $t>t_0$; Theorem 1.1 is false exactly in that case. Equivalently, find a solution of the linear parabolic equation in the coefficient class arising from the difference-of-graphs construction whose nodal set fails the $H^{n-1}$ bound of Theorem 2.1; that single counterexample removes the foundation.
Extended reading notes
Core claim
On the paper's own terms, the discovery is a complete trichotomy for a compact level set flow $M_t$ with finitely many singularities (Theorem 1.3): the statements 'the flow is non-fattening', 'the inner and outer flows coincide with $M_t$', and 'for every smooth closed MCF $N_t$ with $N_t\not\subseteq M_t$, the function $t\mapsto \dim(M_t\cap N_t)$ is non-increasing' are mutually equivalent. The smooth foundation is Theorem 1.1: for two properly embedded smooth MCFs with one compact, $H^{n-1}(M_t\cap N_t)<\infty$ for $t>0$, dimension is non-increasing, and $H^{n-1}(M_s\cap N_s)=0$ at some $s$ forces $M_t\cap N_t=\emptyset$ for all later times, so there is a threshold time $t_0$ separating a phase of positive codimension-two measure from a disjoint phase. Theorem 1.2 gives the identical statement for the self-intersection set of a closed immersed flow without assuming the image is embedded.
Load-bearing premise
The load-bearing premise is the quoted nodal-set estimate for linear parabolic PDEs with uniformly elliptic, Lipschitz coefficients: the difference of two graphical mean curvature flows must satisfy exactly that theorem's hypotheses, and if any edge case escapes the estimate, the finiteness of $H^{n-1}(M_t\cap N_t)$ and all dimension monotonicity built on it collapse.
Editorial extensions
If this is right
- Smooth mean curvature flows obey a sharp higher-dimensional analogue of the one-dimensional curve-shortening intersection-counting theorem: the codimension-two measure is finite for positive times and only the Hausdorff dimension is monotone.
- A closed immersed mean curvature flow becomes an embedding instantaneously at the moment its self-intersection set has $(n-1)$-dimensional measure zero, and it then stays embedded.
- For a level set flow with finitely many singularities, verifying non-fattening does not require knowing the whole flow: it suffices to check that the dimension of intersection with every smooth closed MCF is non-increasing, and a single sphere with an $(n-1)$-dimensional intersection that later meets the flow detects fattening.
- Non-fattening level set flows with finitely many singularities coincide with their inner and outer flows, a statement previously conjectured for all non-fattening flows and now proved in this case via the equivalence.
- The dimension-monotonicity theorem for weak solutions is tight: fattening conical singularities produce Brakke flows whose intersection with a plane increases in dimension, so some restriction like localizability or finiteness of singularities is genuinely necessary.
Reading between the lines
- If the finite-singularity assumption can be relaxed to localizability, the intersection principle becomes a plausible general characterization of non-fattening; the paper's Theorem 4.22 already gives such a statement modulo the multiplicity-one conjecture.
- The sphere-intersection criterion suggests a numerical fattening detector: run a level set computation, intersect with a family of round spheres through the suspected singularity, and look for a jump in intersection dimension; a dimension increase should flag fattening.
- Because the proof scheme only needs a difference function satisfying a linear parabolic equation with controlled coefficients, the same intersection principle should hold for any geometric flow whose graphical evolution has that structure, and in ambient manifolds with bounded geometry, as the paper notes as an expectation.
- A likely sharper picture is that the right monotone quantity in higher codimension is Hausdorff dimension, while the $(n-1)$-measure and component counts are provably unstable; future non-fattening criteria should therefore be phrased in dimension terms.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper develops an 'intersection principle' for mean curvature flow, generalizing the avoidance principle to flows that intersect. Theorem 1.1 shows that for two properly embedded smooth MCFs, one compact, the Hausdorff dimension of their intersection is non-increasing in time and the codimension-two measure is finite after the initial time, with a sharp dichotomy between a positive-measure intersection phase and a disjoint phase. Theorem 1.2 extends the dimension-monotonicity statement to the self-intersection set of a closed immersed MCF, and Proposition 3.7 proves that a small self-intersection set (vanishing (n-1)-measure) forces the immersion to be instantaneously embedded. For weak solutions, the paper introduces a 'localizability' condition for Brakke flows and level set flows, proves intersection dimension monotonicity for localizable flows under a no-higher-multiplicity planar tangent flow assumption (Theorem 4.7 and Theorem 4.22), proves that non-fattening level set flows with finitely many singularities are non-discrepant (Theorem 4.37), and then states the main equivalence Theorem 1.3: for a compact level set flow with finitely many singularities, non-fattening, agreement of inner/outer flows, and the intersection principle with respect to smooth flows are equivalent.
Significance. If the results are correct, the paper gives the first higher-dimensional analogue of Angenent's Sturmian intersection theory for MCF, with sharp dimension-monotonicity statements and a characterization of non-fattening for level set flows with finitely many singularities. The main tools are used carefully: the difference of graphical flows is written as a linear parabolic equation with Lipschitz coefficients (Proposition 2.3), the Huang-Jiang nodal estimate is applied with hypotheses verified by Ecker-Huisken estimates, and the monotonicity strategy through one-sidedness and the strong maximum principle is transparent. The examples in Section 5 are a valuable contribution, showing that measure and component monotonicity fail for n>1 and that dimension monotonicity fails for general Brakke flows. The paper has no fitted parameters and states its hypotheses precisely. However, the proof of the central equivalence (1)⇒(3) in Theorem 1.3 contains a gap: the argument that the no-higher-multiplicity-planar-tangent-flow assumption can be dropped does not, as written, cover intersection points at later singular times.
major comments (2)
- [Section 4.4, proof of Theorem 1.3, final paragraph] The proof that the no-higher-multiplicity-planar-tangent-flow assumption can be dropped is incomplete. In Theorem 4.7 the contradiction proving (4.12) must work for an arbitrary t*>t_i, but the manuscript's replacement argument only establishes smoothness of M^{1,i}_t and M^{2,i}_t on an interval (t0,t0+δ) after the singular time t0. If t* is a later singular time, the subflows are not smooth at t*, and finitely many singularities alone does not imply that their tangent-flow multiplicities k1,k2 are both one, since the multiplicity-one conjecture is open in general. Thus the unconditional implication (1)⇒(3) in Theorem 1.3 is not justified by the text as written. A repair would be to prove emptiness of the intersection on (t0,t0+δ) and then invoke the avoidance principle for weak set flows from t0+δ/2 onward; this step should be stated explicitly.
- [Section 4.4, proof of Theorem 1.3, final paragraph] Even inside the smooth interval (t0,t0+δ), the replacement of Case 2 of Theorem 4.7 is only sketched. If x lies in the intersection of the supports of both subflows and on N_{t*}, then the strong maximum principle implies that each subflow coincides with N_{t*} in a neighborhood; the sum of the two flows then has density two at that point, contradicting unit regularity and the absence of singularities in the interval. The manuscript's sentence that "the strong maximum principle for smooth flows can obviate the need for the multiplicity assumption in the proof of Case 2" should be expanded into a precise argument, since the Case 2 contradiction in Theorem 4.7 previously relied exactly on the no-higher-multiplicity assumption to rule out k1+k2>1.
minor comments (5)
- [Theorem 4.7, statement] There is a typo in the second clause of condition (2): "Moreover, if ifNt is any smooth closed connected MCF and ifdim(...)" should read "Moreover, if N_t is any smooth closed connected MCF and if dim(...)".
- [Definition 4.5] The notation "t0−δ2,t0+δ2" appears without parentheses and is ambiguous; it should be written as (t0−δ^2, t0+δ^2) or the intended interval should be spelled out.
- [Proof of Theorem 4.7, after equation (4.12)] The argument that K_t is the level set flow of the initial closed domain K_{t_i} is used implicitly when applying Lemma 4.3. This is a standard fact for smooth flows of hypersurfaces bounding domains, but it should be stated or cited explicitly, since the containment (spt M^{1,i})_t ⊆ K_t is load-bearing for the rest of the proof.
- [Section 2.1, Theorem 2.10] In the first paragraph of the proof, the statement that equality of Mt and Nt in a ball at fixed t would imply equality in U for all t>0 is compressed; the spatial identity theorem alone gives equality at that time, and one needs continuity/smoothness in time or a separate argument to conclude equality of the flows. This is a minor presentation issue, but the implication should be clarified.
- [Section 5.1, Corollary 5.2] The application of Theorem 1.2 to an ancient self-shrinking flow requires translating the time interval; the proof should state that one applies Theorem 1.2 on intervals of the form [-1,0) and uses scaling, since Theorem 1.2 is stated for flows on [0,T).
Circularity Check
No significant circularity: main theorems are derived from external results (Huang–Jiang, White, Hershkovits–White, Solomon–White) and the few author self-citations are independent prior statements, not fitted or self-defined inputs.
full rationale
The paper's derivation chain is self-contained in the relevant sense. Theorem 1.1 is built on the external Huang–Jiang nodal estimate (Theorem 2.1, quoted from [HJ24, Theorem 1.4]), the Ecker–Huisken interior estimates, and standard maximum-principle arguments; no quantity entering the theorem is fitted to the claimed conclusion, and the H^{n-1}-finiteness bound is a consequence of the linear parabolic nodal estimate rather than an input. Theorems 1.2 and 1.3 are likewise deduced from stated hypotheses together with cited external results: White's topological-change theorem, Hershkovits–White's inner/outer flow construction, Solomon–White's strong maximum principle for varifolds, and Ilmanen's weak set flow comparison. The author self-citations that occur ([Pay20, Proposition 3.4] and [MP21]) are used as prior mathematical statements, not as assumptions equivalent to the target theorems: [Pay20] supplies a support-identification statement for boundary motions, and [MP21] supplies an eternal solution used in a counterexample in Section 5. No step renames an input as a prediction, fits a parameter to a subset of data and then 'predicts' a closely related quantity, or imports a uniqueness theorem from the authors' own prior work to force the main choice. The principal risk in the proof of Theorem 1.3—the passage in Section 4.4 claiming that 'in the case of finitely many singularities, we can find δ>0 such that both M^{1,i}_t and M^{2,i}_t are globally smooth at time t∈(t0,t0+δ)' while (4.12) is asserted for arbitrary t*>t_i—is a potential correctness gap concerning later singular times, not a circularity: the argument does not define non-fattening, non-discrepancy, or the intersection principle in terms of themselves. Therefore the circularity score is 0.
Assumptions & free parameters
assumptions (7)
- standard math Huang-Jiang nodal set estimate for linear parabolic PDEs (Theorem 2.1, [HJ24, Theorem 1.4])
- standard math Backward uniqueness for complete smooth mean curvature flows with bounded second fundamental form ([Hua19])
- standard math Ecker-Huisken interior gradient and curvature estimates ([EH91])
- standard math White's theorem on topological changes of level set flow ([Whi95, Theorem 1(i)])
- standard math Hershkovits-White inner and outer flow properties ([HW20, Theorem B.2, Appendix B])
- standard math Solomon-White strong maximum principle for stationary varifolds ([SW89])
- domain assumption Standing assumption: every Brakke flow considered has bounded area ratios
Cite this review
Pith. "Pith review of An Intersection Principle for Mean Curvature Flow." pith.science (2026). https://pith.science/paper/IA2IV5ZN
@misc{pith2026250511600,
author = {Pith},
title = {Pith review of: An Intersection Principle for Mean Curvature Flow},
year = {2026},
howpublished = {\url{https://pith.science/paper/IA2IV5ZN}},
note = {Machine review of arXiv:2505.11600}
}
read the original abstract
The avoidance principle says that mean curvature flows of hypersurfaces remain disjoint if they are disjoint at the initial time. We prove several generalizations of the avoidance principle that allow for intersections of hypersurfaces. First, we prove that the Hausdorff dimension of the intersection of two mean curvature flows is non-increasing over time, and we find precise information on how the dimension changes. We then show that the self-intersection of an immersed mean curvature flow has non-increasing dimension over time. Next, we extend the intersection dimension monotonicity to Brakke flows and level set flows which satisfy a localizability condition, and we provide examples showing that the monotonicity fails for general weak solutions. We find a localization result for level set flows with finitely many singularities, and as a consequence, we obtain a fattening criterion for these flows which depends on the behavior of intersections with smooth flows.
Forward citations
Cited by 1 Pith paper
-
An Ancient Stacked Pancake Solution to Mean Curvature Flow
For every dimension n≥3, there exists an embedded, rotationally symmetric, non-convex ancient mean curvature flow that looks like two parallel pancakes joined by a neck and lies in a slab.
Reference graph
Works this paper leans on
-
[1]
[AAG95] S. Altschuler, S. B. Angenent, and Y . Giga, Mean curvature flow through singularities for surfaces of rotation, J. Geom. Anal. 5 (1995), no. 3, 293–358. [ACI95] S. B. Angenent, D. Chopp, and T. Ilmanen, A computed example of nonuniqueness of mean curvature flow in R3, Commun. Partial Differ. Equ. 20 (1995), no. 11-12, 1937–1958. [AIV17] S. Angene...
work page 1995
-
[3]
Proceedings from the third conference, held August 20-29, 1989 in Gregynog, Wales, United Kingdom, Boston, MA etc.: Birkh¨auser, 1992, pp. 21–38. [BK23] R. Bamler and B. Kleiner, On the multiplicity one conjecture for mean curvature flows of surfaces, arXiv:2312.02106,
arXiv 1989
-
[12]
[Mit20] B. S. Mityagin, The zero set of a real analytic function, Mat. Zametki 107 (2020), no. 3, 473–475. MR 4070868 [MP21] A. Mramor and A. Payne, Ancient and eternal solutions to mean curvature flow from minimal surfaces , Math. Ann. 380 (2021), no. 1-2, 569–591. MR 4263692 [Pay20] A. Payne, Mass drop and multiplicity in mean curvature flow, arXiv:2009.14163,
work page Pith review arXiv 2020
-
[32]
Ketover, Self-shrinkers whose asymptotic cones fatten, arXiv:2407.01240,
MR 4888468 [Ket24] D. Ketover, Self-shrinkers whose asymptotic cones fatten, arXiv:2407.01240,
-
[111]
AN INTERSECTION PRINCIPLE FOR MEAN CURV ATURE FLOW 39 [Wic14] N. Wickramasekera, A sharp strong maximum principle and a sharp unique continuation theorem for singular minimal hypersurfaces, Calc. Var. Partial Differ Equ.51 (2014), no. 3-4, 799–812. MR 3268871 MASSACHUSETTS INSTITUTE OF TECHNOLOGY , 77 M ASSACHUSETTS AVENUE , CAMBRIDGE , MA 02139, USA Emai...
work page 2014
-
[1990]
54 (1993), 193–204. [Ilm94] , Elliptic regularization and partial regularity for motion by mean curvature, Memoirs of the American Math- ematical Society 108 (1994), no. 520, 1–90. [Ilm95] , Lectures on mean curvature flow and related equations , Notes, http://people.math.ethz.ch/ ˜ilmanen/papers/notes.pdf,
work page 1993
-
[1995]
[INS19] T. Ilmanen, A. Neves, and F. Schulze, On short time existence for the planar network flow, J. Differ. Geom.111 (2019), no. 1, 39–89. [IW25] T. Ilmanen and B. White, Fattening in mean curvature flow, Ars Inven. Anal. (2025), Paper No. 4,
work page 2019
- [1996]
Show all 16 references
-
[2010]
Chodosh and F
[CS21] O. Chodosh and F. Schulze, Uniqueness of asymptotically conical tangent flows , Duke Math. J. 170 (2021), no. 16, 3601–3657. MR 4332673 AN INTERSECTION PRINCIPLE FOR MEAN CURV ATURE FLOW 38 [CY07] B.-L. Chen and L. Yin, Uniqueness and pseudolocality theorems of the mean...
2021 arXiv
-
[2011]
[CMP15] T. H. Colding, W. P. Minicozzi, II, and E. K. Pedersen, Mean curvature flow, Bull. Amer. Math. Soc. (N.S.) 52 (2015), no. 2, 297–333. MR 3312634 [Coo10] A. A. Cooper, A compactness theorem for the second fundamental form., arXiv:1006.5697,
2015 arXiv
-
[2017]
Andrews, M
[ALM13] B. Andrews, M. Langford, and J. McCoy, Non-collapsing in fully non-linear curvature flows , Ann. Inst. H. Poincar ´e Anal. Non Lin´eaire 30 (2013), no. 1, 23–32. MR 3011290 [Ang88] S. Angenent, The zero set of a solution of a parabolic equation, J. Reine Angew. Math.39...
2013
-
[2020]
Peachey, Non-uniqueness of curve shortening flow, arXiv:2205.03442,
[Pea22] L. Peachey, Non-uniqueness of curve shortening flow, arXiv:2205.03442,
-
[2022]
Solomon and B
[SW89] B. Solomon and B. White, A strong maximum principle for varifolds that are stationary with respect to even parametric elliptic functionals, Indiana Univ. Math. J. 38 (1989), no. 3, 683–691. MR 1017330 [SWX25] A. Sun, Z. Wang, and J. Xue, Passing through nondegenerate si...
1989 arXiv
-
[2023]
Bourni, M
[BLT21] T. Bourni, M. Langford, and G. Tinaglia, Collapsing ancient solutions of mean curvature flow , J. Differ. Geom. 119 (2021), no. 2, 187–219. [CCMS24] O. Chodosh, K. Choi, C. Mantoulidis, and F. Schulze, Mean curvature flow with generic initial data, Invent. Math. 237 (2...
2021
-
[2024]
Han and F.-H
[HL94] Q. Han and F.-H. Lin, Nodal sets of solutions of parabolic equations: II, Commun. Pure Appl. Math. 47 (1994), no. 9, 1219–1238. [Hua19] H. Huang, Backwards uniqueness of the mean curvature flow, Geom. Dedicata 203 (2019), 67–71. [HW20] O. Hershkovits and B. White, Nonfa...
1994
-
[2025]
White, The topology of hypersurfaces moving by mean curvature, Comm
[Whi95] B. White, The topology of hypersurfaces moving by mean curvature, Comm. Anal. Geom. 3 (1995), no. 1-2, 317–333. MR 1362655 [Whi97] , Stratification of minimal surfaces, mean curvature flows, and harmonic maps. , J. Reine Angew. Math. 1997 (1997), no. 488, 1–36. [Whi00]...
1995
Reviewed August 15, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.