{"id":"8e19df29-2588-4530-bbe8-26d9b9c34fd3","arxiv_id":"2501.16678","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Mean curvature flows through nondegenerate cylindrical singularities undergo an isolated, graphical surgery event whose topology change equals an (n-k)-surgery, matching Morse level sets.","lead":"A new proof shows that the only way a mean curvature flow can pass through a nondegenerate cylindrical singularity is a clean pinching-and-surgery event: the singularity is isolated, the surface stays graphical around it, and the topology change matches a Morse-theoretic handle attachment. This gives a canonical description of surgery for all dimensions of the Euclidean factor, a step toward using mean curvature flow to extract topological information from hypersurfaces.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 5.3(c) postulates an unproved smooth boundary deformation that is essential for forward-time noncollapsing and for applying the Du-Zhu classification; without it, items (i) and (vii) of Theorem 1.1 are not established.","rationale":"I read the paper in good faith. The central theorem is a substantial and coherent extension of [SX22] and the new decay-order machinery is an original tool. However, the forward-time part of Theorem 1.1 rests on Proposition 5.3, and that proposition contains an unproved existence hypothesis (c). The reader correctly identified this as the weakest point. The deformation Γ_t is not a cosmetic technicality: it is exactly what makes the elliptic-regularization minimizers have a controlled boundary at the top of the cylinder, which is what yields the two-sided noncollapsing bounds via the maximum principle. Without noncollapsing, the blow-up limits in Theorem 4.2 cannot be fed into the Du-Zhu classification, so the forward-time isolation and the graphical representation over the dual cylinder would not follow. The paper explicitly flags (c) as conjecturally removable but does not remove it or prove it. I also note a mismatch between the boundary control in Theorem 1.1(iv) on ∂Q_{r0} and the boundary in Proposition 5.3(c) on ∂B_{r0}, which makes the 'easy to check' assertion less obvious than stated. I found no other single gap of comparable weight: the reliance on [SX22] is acceptable as an independent prior theorem, and the long L2 estimates, though long, are laid out in detail. If the deformation is supplied (or removed with a proof), the paper should be accepted; otherwise it is conditional.","tokens_in":52072,"tokens_out":10520,"duration_ms":94028,"concrete_test":"Construct explicitly, for the model nondegenerate C_{n,k} singularity (or for the flow in Theorem 1.1 after choosing r0), the smooth monotone family Γ_t on ∂B_{r0} required by Proposition 5.3(c), using the boundary evolution (iv) and cusp profile (vi). If no such family can be written down or proved to exist, Proposition 5.3(iii) fails and the forward-time classification collapses; a successful construction would remove the gap.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Proposition 5.3 is the only route to forward-time mean convexity and noncollapsing (Theorem 1.1(ii),(iii)). Its condition (c) requires a smooth monotonic deformation Γ_t on ∂B_{r0} with Γ_t = spt M_t∩∂B_{r0} for t≤T and Γ_t = (S^{n-k}(r0')×R^k)∩∂B_{r0} for t≥T+1. This deformation is used in the elliptic-regularization proof of (iii): it defines the boundary portion of the minimizers N_λ, and the smoothness/monotonicity of Γ_t at the top and side boundaries is what yields the uniform two-sided bounds on Z*/H and Z_*/H that pass to the limit as λ→∞. The proof of Proposition 5.3 never constructs Γ_t for the nondegenerate-singularity flow of Theorem 1.1; it only says the proposition 'implies' (ii),(iii) using (iv)-(vi), and adds a conjecture that (c) can be dropped. This is an omitted proof of a hypothesis that is load-bearing: without the noncollapsing conclusion, Theorem 4.2 cannot invoke the Du-Zhu classification (Theorem 2.10), so the forward-time isolatedness (i) and the graphical/morphism description (vii) are unsupported. Moreover, matching the boundary on ∂B_{r0} is nontrivial: Theorem 1.1(iv) controls the flow on ∂Q_{r0}, not on the Euclidean sphere ∂B_{r0}, so the 'easy to check' claim is not immediate.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies unit-regular cyclic mod 2 Brakke flows with a nondegenerate cylindrical singularity modeled on C_{n,k} at a point. The main result, Theorem 1.1, claims that such a singularity is isolated in a parabolic neighborhood, that the flow is mean convex and noncollapsing there, that the flow is graphical over the cylinder before and at the singular time (with a cusp profile |y|/(2 sqrt(-log|y|)) at the singular time), that after the singular time it is a graph over the dual cylinder, and that the topology change is an (n-k)-surgery. The proof introduces an L^2-distance monotonicity formula and a discrete decay order, and combines them with pseudolocality, elliptic regularization, White regularity, and the Du-Zhu classification of noncollapsing ancient flows. The backward-time statements are taken from prior work of the first and third authors, and the forward-time statements are derived from a new classification of blow-up models.","tokens_in":52421,"tokens_out":6794,"duration_ms":59941,"significance":"If the main theorem holds, the paper provides a complete local description of the geometry and topology change near nondegenerate cylindrical singularities for arbitrary n and k, including the mean-convex-neighborhood property and surgery description. The new L^2-distance monotonicity (Corollary 3.3) and the discrete almost-monotonicity of the decay order (Corollary 3.7) are original tools with potential applications to other singularities. The paper also derives several global corollaries (Corollaries 1.3-1.8) about uniqueness, finiteness, and handle decompositions under the generic-singularity assumption. The analytic arguments are detailed and use standard machinery; the appendices provide useful technical lemmas. No machine-checked proofs are included, but the derivations are presented in a verifiable manner.","major_comments":[{"comment":"Proposition 5.3(c) postulates a smooth monotonic boundary deformation {Gamma_t}_{t>=0} with Gamma_t = spt M_t cap partial B_{r0} for t in [0,T] and Gamma_t = (S^{n-k}(r0') x R^k) cap partial B_{r0} for t >= T+1, and the paper asserts that this deformation exists and is 'easy to check', while also conjecturing that it can be dropped. No construction of Gamma_t is given for the specific nondegenerate-singularity flow of Theorem 1.1. This hypothesis is load-bearing: it defines the boundary portion of the minimizers N_lambda in the elliptic-regularization proof of Proposition 5.3(iii), and the smoothness/monotonicity of Gamma_t at the top and side boundaries is what yields the uniform two-sided bounds on Z*/H and Z_*/H that pass to the limit as lambda -> infinity. Without the noncollapsing conclusion (iii), Theorem 4.2 cannot invoke the Du-Zhu classification (Theorem 2.10), so the forward-time isolatedness (i) and the graphical/morphism description (vii) of Theorem 1.1 are unsupported.","section":"Section 5.2, Proposition 5.3(c)"},{"comment":"The claimed deformation Gamma_t is not immediate from the results proved earlier. Theorem 1.1(iv) controls spt M(t) cap partial Q_{r0} (the boundary of the product cylinder Q_{r0} = B^{n-k+1}_{r0} x B^k_{r0}), whereas Proposition 5.3(c) requires boundary data on the Euclidean sphere partial B_{r0} cap Omega. These are different boundaries, and no argument is given for how the flow's known boundary behavior on partial Q_{r0} yields a smooth monotone family Gamma_t on partial B_{r0} connecting the pre-singular boundary data to the dual-cylinder cross-section. Until such a construction is supplied, the forward-time noncollapsing conclusion (Theorem 1.1(iii)) and the classification of blow-up models in Theorem 4.2 remain conditional on an unproved assumption.","section":"Section 5.2, Proposition 5.3(c); Theorem 1.1(iv)"}],"minor_comments":[{"comment":"The text contains numerous OCR-type artifacts in the displayed text and references (e.g., 'P ASSING', 'V A TURE', 'W ANG', 'f ˜A¼r'), which should be cleaned before publication.","section":"Throughout"},{"comment":"The phrase 'tubular neighborhood U(t) of {0} x S^{k-1}_{r0} in partial Q_{r0} cap C^*_{n,k}(r0)' is ambiguous because partial Q_{r0} cap C^*_{n,k}(r0) is not a manifold with boundary; please clarify the intended topology.","section":"Theorem 1.1(iv)"},{"comment":"The paper's own remark after Proposition 5.3 acknowledging that assumption (c) is conjectural should be highlighted in the introduction as a limitation, since it currently appears only in the middle of the proof.","section":"Section 5.2, Proposition 5.3"}],"recommendation":"major_revision","confidential_remarks":"The main gap is the unproved boundary deformation in Proposition 5.3(c). I recommend the editor ask the authors to either construct Gamma_t for nondegenerate-singularity flows or prove Proposition 5.3(iii) by a different argument. Given the paper's length and the depth of the remaining analysis, this is a substantial revision rather than a minor edit."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Plainly: this is a serious piece of work with a genuinely new main theorem, but there is one unproved hypothesis in the proof of forward-time noncollapsing that the authors need to supply or remove. The reader's conditional verdict is right, and the stress-test note lands on the right spot.\n\nWhat is new: Theorem 1.1 is the first complete local description of a flow passing through a nondegenerate cylindrical singularity for all 1 <= k <= n-1, not just k=1 or C_{4,3}. The L^2-distance monotonicity formula and the discrete decay order are real analytical tools, and they are used to get isolatedness, the graphical descriptions before and after the singular time, and the surgery interpretation. I do not see circularity in relying on [SX22]: that is an independent prior theorem, and the new decay-order argument is self-contained.\n\nThe proof structure is coherent: pseudolocality, elliptic regularization, White regularity, and the Du-Zhu classification are appropriate machinery. I could not find an obvious contradiction in the central analytic estimates, though I would not certify every line of the long computations without more time.\n\nThe soft spot is Proposition 5.3(c). The paper postulates a smooth monotonic deformation of boundary data on the Euclidean sphere and calls it 'easy to check'; it is not immediate, since the control from Theorem 1.1(iv) is on the boundary of the parabolic box Q_{r0}, not on the sphere. This deformation is load-bearing: through the elliptic-regularization argument it yields the two-sided noncollapsing bound, and without noncollapsing the forward-time classification via Du-Zhu (Theorem 4.2) is unavailable. The authors themselves state that they conjecture (c) can be dropped; that is honest, but it means the forward-time conclusions, in particular items (i), (vii), and (viii), are not fully established as written.\n\nA minor point: the topological corollaries are all conditional on assumption (⋆) and many are also conditional on Conjecture 1.4. They are fine as indications of what the theorem is good for, but they should not be read as unconditional applications.\n\nBottom line: this paper deserves a serious referee. The conditional verdict is fair. If the deformation assumption can be supplied, or a different route to noncollapsing is found, this is a strong paper. I would send it to a good journal and ask the authors to address Proposition 5.3(c) in a revision. I would not cite it in my own work in the next year (not my area), but I would bring it to the reading group.","headline":"A substantial new theorem about MCF through nondegenerate cylindrical singularities, but one unproved boundary-deformation hypothesis in Proposition 5.3(c) is load-bearing for the forward-time conclusions.","tokens_in":52953,"tokens_out":1609,"would_cite":false,"duration_ms":17415,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53E10","35K93","57R70"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a nondegenerate cylindrical singularity of a mean curvature flow is isolated in spacetime, that the surrounding flow is mean convex, noncollapsing, and a smooth graph just before and just after the singular instant…","keywords":["mean curvature flow","nondegenerate cylindrical singularity","Brakke flow","neck pinch","decay order","frequency function","Morse theory","noncollapsing"],"falsifier":"Take a concrete nondegenerate neck pinch, such as the rotationally symmetric examples of [AV97], track the boundary curves $\\mathrm{spt}\\,M(t)\\cap\\partial B_{r_0}$ through the singular instant, and check whether a smooth monotonic family $\\Gamma_t$ joining them to the dual-cylinder slices $(S^{n-k}(r_0')\\times\\mathbb{R}^k)\\cap\\partial B_{r_0}$ exists; if the boundary data folds back on itself or cannot be joined monotonically, the hypothesis of Proposition 5.3(c) fails and the noncollapsing argument — and with it the forward-time classification — does not go through.","tokens_in":51858,"feed_emoji":"✂️","tokens_out":16066,"duration_ms":127586,"temperature":0.7,"pith_summary":"The paper aims to give a complete local answer to what happens when a mean curvature flow passes through a nondegenerate cylindrical singularity, the generic neck pinch in which a hypersurface collapses onto a generalized cylinder $C_{n,k}=S^{n-k}(\\sqrt{2(n-k)})\\times \\mathbb{R}^k$. It claims that such a singularity is isolated in spacetime, that the flow near it is mean convex and noncollapsing, and that the moving surface is a smooth graph over the cylinder before the singular time, a cusped graph at the singular time, and a smooth graph over the dual cylinder afterwards. It further claims that the topology change is exactly the change of level sets near a Morse critical point, namely an $(n-k)$-surgery. Because the result holds for every dimension $n\\ge 2$ and every cylinder type $C_{n,k}$, it supplies a canonical, parameter-free picture of the surgery that a mean curvature flow performs at a neck pinch.","feed_headline":"Neck pinch acts exactly like Morse surgery, in every dimension","feed_subtitle":"Nondegenerate neck pinches are isolated; the flow is a graph before, a cusp at, a graph after.","key_machinery":"The engine is a new weighted $L^2$-distance monotonicity on the rescaled flow. Define $d_{n,k}(\\Sigma)^2=\\int \\mathrm{dist}_{n,k}(X)^2 e^{-|X|^2/4}\\,d\\Sigma$, where $\\mathrm{dist}_{n,k}$ is a regularized signed distance to $C_{n,k}$; a non-concentration estimate shows that the far-away part of this integral is controlled by its initial value. From it the paper defines the decay order $N_{n,k}(\\tau;M)=\\log\\big(d_{n,k}(M(\\tau))/d_{n,k}(M(\\tau+1))\\big)$, a discrete analogue of Almgren's frequency function, and proves a discrete almost-monotonicity: the decay order either drops by a definite amount or stabilizes near an eigenvalue of the Jacobi operator $-L_{n,k}$ on the cylinder. Since the small eigenvalues of $-L_{n,k}$ correspond exactly to the unstable modes (translation in the sphere and in the $\\mathbb{R}^k$ directions, plus the quadratic Hermite modes), the stabilization identifies which linear mode dominates the flow. That identification, combined with a classification of noncollapsing ancient asymptotically cylindrical flows as either shrinking cylinders or bowl solitons times $\\mathbb{R}^{k-1}$, yields the dichotomy of blow-up models from which the isolatedness conclusion and the post-singular graphical description follow.","core_discovery":"The central claim is Theorem 1.1. For a unit-regular cyclic mod 2 Brakke flow $t\\mapsto M(t)$ in $\\mathbb{R}^{n+1}$ with a nondegenerate cylindrical singularity modeled by $C_{n,k}$ at the spacetime origin, the paper asserts that $(0,0)$ is the only singularity in a parabolic neighborhood $Q_{r_0}\\times[-t_0,t_0]$; that the flow there is mean convex and noncollapsing; that for $t<0$ the surface is a $C^\\infty$ graph over $C_{n,k}$; that at $t=0$ it is a graph with the universal cusp profile $u(\\theta,y)=\\sqrt{2(n-k)}\\,\\frac{|y|}{2\\sqrt{-\\log|y|}}\\,(1+o_y(1))-\\sqrt{2(n-k)}$; that for $t>0$ it is a smooth graph over the dual cylinder $C^*_{n,k}(r_0)=\\mathbb{R}^{n-k+1}\\times S^{k-1}(r_0)$; and that the topology change is an $(n-k)$-surgery, identical to the level-set transition near a Morse critical point of index $n-k+1$.","pith_inferences":["If the paper's Conjecture 1.4 is settled, the theorem becomes a 'missing handle' principle: the spacetime track of a generic mean convex flow would have a handle decomposition with no $n$-handles or $(n+1)$-handles, so standard balls would be the only building blocks.","The decay-order machinery is not restricted to nondegenerate singularities, and the paper announces a companion study of degenerate ones; a concrete test is to compute $N_{n,k}(\\tau)$ in the rotationally symmetric neck-pinch examples of [AV97], where the cusp profile is explicit, and check that the limiting decay order lands on the predicted eigenvalue.","The cusp profile yields a quantitative prediction that numerical simulation could check: at the singular instant the neck radius should close like $|y|/(2\\sqrt{-\\log|y|})$ as a function of distance from the spine, independently of the initial shape.","Isolatedness of each nondegenerate singularity is exactly the local input needed to run surgeries sequentially, so a suitable a priori bound on the number of singular events could turn the local description into a global decomposition of any flow satisfying the nondegeneracy condition."],"forward_implications":["Corollary 1.3: a mean curvature flow whose only singularities are nondegenerate cylindrical and spherical ones is unique, has only finitely many singularities, and its spacetime track admits a Morse function whose index-$(n-k+1)$ critical points are in one-to-one correspondence with the singularities modeled by $C_{n,k}$.","Corollary 1.5: if the flow starts from a closed $k$-convex hypersurface and satisfies the nondegeneracy condition, the enclosed domain carries a Morse function with no critical points of index $0,1,\\dots,n-k+1$, so the domain is obtained from standard balls by attaching only handles of indices $1$ through $k-1$.","Corollaries 1.7 and 1.8: the Betti numbers of the initial hypersurface force a lower bound on the number of nondegenerate singularities of each cylinder type, because each topology change is now a completely understood surgery.","The singular-time profile is universal: at the singular instant the surface is a graph over $C_{n,k}$ whose deviation from the cylinder is $\\sqrt{2(n-k)}\\,|y|/(2\\sqrt{-\\log|y|})$ near the spine, with no dependence on the particular flow.","The surgery performed by the flow is canonical: the graphical descriptions before and after the singular time leave no freedom to choose when or where to cut, unlike earlier surgery constructions."],"supporting_citations":[{"why":"supplies the normal-form theorem for nondegenerate cylindrical singularities, the graphical description before the singular time, and the pseudolocality-based items (iv)-(vi).","marker":"[SX22]"},{"why":"classification of noncollapsing ancient asymptotically cylindrical flows into shrinking cylinders and bowl solitons times R^{k-1}, the dichotomy used in the blow-up classification Theorem 4.2.","marker":"[DZ22]"},{"why":"Brakke-White epsilon regularity that converts the blow-up classification into the statement that the singularity is isolated.","marker":"[Whi05]"},{"why":"Huisken's monotonicity formula, which defines the Gaussian density and the rescaled-flow setup on which the L2-distance analysis is built.","marker":"[Hui90]"},{"why":"elliptic regularization with prescribed boundary data that Proposition 5.3 uses to prove mean convexity and noncollapsing.","marker":"[Whi15]"},{"why":"uniqueness and rigidity of blow-ups at cylindrical singularities, used to upgrade rescaled convergence to smooth graphical convergence.","marker":"[CM15]"},{"why":"nonfattening criterion for singularities of mean convex type used to derive uniqueness in Corollary 1.3.","marker":"[HW20]"},{"why":"the rotationally symmetric analysis that produced the degenerate neck-pinch examples and the cusp asymptotics the paper generalizes in item (vi).","marker":"[AV97]"}],"fun_headline_variants":["Neck pinch = Morse surgery, in every dimension","Isolated neck pinches: flow does Morse surgery","Cusp at neck pinch, then graph: surgery like Morse","Mean curvature flow: neck pinch is isolated Morse surgery","Nondegenerate neck pinch: cusp profile, Morse-like topology change"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof that the flow is noncollapsing after the singular time, and through that the classification of all possible blow-up shapes, assumes that the boundary traces of the flow on a fixed sphere can be smoothly and monotonically deformed into the boundary traces of the dual cylinder; the paper calls this “easy to check” while conjecturing it can be dropped, and if no such deformation exists the noncollapsing step is unsupported.","fun_headline_variants_meta":{"raw":{"variants":["Neck pinch = Morse surgery, in every dimension","Isolated neck pinches: flow does Morse surgery","Cusp at neck pinch, then graph: surgery like Morse","Mean curvature flow: neck pinch is isolated Morse surgery","Nondegenerate neck pinch: cusp profile, Morse-like topology change"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000219,"raw_usage":{"total_tokens":1426,"prompt_tokens":909,"completion_tokens":517,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":525,"completion_tokens_details":{"reasoning_tokens":435}},"tokens_in":525,"tokens_out":517,"duration_ms":5286,"temperature":1.0,"reasoning_tokens":435,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T11:28:53.988796+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a concrete nondegenerate neck pinch, such as the rotationally symmetric examples of [AV97], track the boundary curves $\\mathrm{spt}\\,M(t)\\cap\\partial B_{r_0}$ through the singular instant, and check whether a smooth monotonic family $\\Gamma_t$ joining them to the dual-cylinder slices $(S^{n-k}(r_0')\\times\\mathbb{R}^k)\\cap\\partial B_{r_0}$ exists; if the boundary data folds back on itself or cannot be joined monotonically, the hypothesis of Proposition 5.3(c) fails and the noncollapsing argument — and with it the forward-time classification — does not go through.","supporting_citations":[],"review_version":1}