{"id":"91521ca7-81bb-4c91-a59b-746824e5dd07","arxiv_id":"2411.19293","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For 5 to 9 dimensions, an infinite-dimensional family of Yang-Mills flow solutions is constructed that converge, modulo gauge, to the homothetically shrinking soliton W, including asymmetric Type-I blowups.","lead":"This paper proves that the Yang-Mills flow on R^n, for dimensions 5 through 9, admits many solutions that blow up at a finite time while converging, up to gauge, to a known shrinking soliton; some of these solutions are not symmetric under SO(n). The result gives the first non-equivariant Type-I blowup examples and shows the soliton is dynamically stable in a much larger, infinite-dimensional sense.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (2.5) does not define a single operator: the two displayed formulas for A=-U^{-1}LU differ (e.g., the [φ,F] coefficient and missing y·[W,φ] terms), so the spectral Lemma 2.1 that underpins Π_{>0} may not apply to the true linearized operator.","rationale":"The Reader's CONDITIONAL verdict is justified. I agree with the weakest_assumption that Lemma 2.1 is foundational. My pass sharpens it into a concrete algebraic inconsistency in the displayed definition of A. This is not a disagreement with consensus nor an external objection; it is an internal correctness risk in the operator identity that the spectral proof uses. If the identity is merely a typo, the paper is likely repairable; if the true A has first-order terms that are not self-adjoint, then the projection operator and all estimates built on it fail. Therefore I recommend keeping CONDITIONAL (no change from the Reader), pending the one-line verification. The remainder of the paper—Kato inequality, barrier estimates, and fixed-point argument—appears coherent, and I found no circularity; the asymmetry argument in Corollary 1.2 is plausible conditional on the spectral setup. I do not raise the sign issue in Lemma 3.9 for positive eigenfunctions as the headline because non-positive eigenfunctions are the ones actually used in (4.2), and their case is covered even if the displayed estimate has a sign slip.","tokens_in":26321,"tokens_out":35939,"duration_ms":314226,"concrete_test":"Re-derive A=-U^{-1}LU from (2.4) by expanding every term: Δ(e^{r^2/8}φ), -(1/2)y·∇(e^{r^2/8}φ), 2[W_i,∂_i(e^{r^2/8}φ)], etc., and compare with both displays in (2.5) for a compactly supported so(n)-valued test 1-form. If the two displays differ, determine the corrected operator; then rerun the compactness argument of Lemma 2.1 for the corrected A. The concern is settled if the corrected A is still a self-adjoint Witten-type Laplacian with compact resolvent; otherwise the spectral basis and Π_{>0} are not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 2.1 is load-bearing: it supplies the complete L^2_ρ eigenbasis, the finite nonpositive spectrum, and the projector Π_{>0} used in Theorem 4.1. Its proof is run not on L itself but on the conjugated operator A defined in (2.5). The two expressions for A displayed in (2.5) are not equal as written. Conjugating the term 2[W_i,∂_i u_j] from (2.4) by U=e^{|·|^2/8} produces (1/2)y_i[W_i,φ_j] in addition to 2[W_i,∂_i φ_j]; no such term appears in either display. The coefficient of [φ_i,F_{W,ij}] changes from 2 in the first display to 1 in the second, and the zeroth-order constants do not match. Since self-adjointness and compact resolvent are asserted for the second display, an unproved equivalence is needed before Lemma 2.1 can be applied to the L that actually appears in (3.1). If the identity fails, the discrete spectral decomposition used to define the solution formula (3.4) and every contraction estimate has no basis, so the central stable-manifold claim is unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs, for 5 ≤ n ≤ 9, an infinite-dimensional family of smooth solutions to the Yang–Mills flow on R^n with structure group SO(n) that converge, modulo gauge, to Weinkove's homothetically shrinking SO(n)-equivariant soliton W, thereby producing asymmetric Type-I blowup solutions. The proof proceeds through the de-Turck gauge, similarity variables, linearization at W, a Gaussian L^2 spectral decomposition of the linearized operator, weighted parabolic Hölder estimates, and a fixed-point argument on a projected stable slice, followed by gauge reconstruction. The main results are Theorem 1.1, Theorem 4.2, and Corollary 1.2.","tokens_in":26497,"tokens_out":31003,"duration_ms":259494,"significance":"If correct, this would be a notable advance: it would give the first construction of non-equivariant Type-I blowup for the Yang–Mills flow in dimensions 5 ≤ n ≤ 9 and would show that the stable manifold of the soliton is infinite-dimensional in the full gauge-fixed problem. The overall strategy is appropriate, and the construction is not circular: the soliton is imported explicitly from Weinkove's work, the linearized equation is derived from the flow, and the contradiction argument for non-equivariance is self-contained. The explicit translation eigenfunctions and the refined Kato inequality are potentially useful contributions. The significance is contingent, however, on correcting the linearization and spectral input discussed in the major comments.","major_comments":[{"comment":"The displayed identities defining L and A are internally inconsistent. In Eq. (2.4), the first expression for L ends with the term 2[u_i, ∂_i W_j − ∂_j W_i + [W_i, W_j]], while the second expression ends with −[u_i, F_{W,ij}] for the same bracket; these differ by a factor of −2, so the equality in (2.4) is false as written. The problem propagates to Eq. (2.5): conjugating the first expression of (2.4) by U = e^{|y|^2/8} produces, from the term 2[W_i, ∂_i u_j], an additional term (1/2)y_i[W_i, φ_j], which is absent from both displayed formulas for A. In addition, the coefficient of [φ_i, F_{W,ij}] changes from 2 in the first display of (2.5) to 1 in the second, and the quadratic potentials differ by an additive constant. Consequently, Lemma 2.1 is proved for an operator that is not shown to be the operator L that appears in Eq. (3.1).","section":"§2.2, Eq. (2.5); §2.1, Eq. (2.4)"},{"comment":"Lemma 2.1 is the load-bearing spectral input of the paper. It supplies the complete L^2_ρ eigenbasis, the finite set of nonpositive eigenvalues, and the spectral gap λ_{I+1} > 0 that are used in the explicit solution formula (3.4), the decay estimate (3.5), Lemma 3.12, and the contraction estimates in Theorem 4.1. Because the operator on which Lemma 2.1 is proved is not unambiguously the true conjugated linearization, the current manuscript does not establish this spectral input. The authors must recompute A for the actual L and prove discreteness and self-adjointness (or a suitable spectral substitute) for that operator. A bounded-perturbation repair may be possible, but it is not supplied; without it the stable-manifold construction has no foundation.","section":"§2.2, Lemma 2.1"}],"minor_comments":[{"comment":"The displayed Hölder manipulations in Eqs. (3.6) and (3.7) appear to contain missing squares and misplaced factors; please rewrite these lines carefully, since they are the source of the exponential decay bounds.","section":"§3.2, Lemma 3.1"},{"comment":"The statement says the convergence estimate holds for all t ∈ [0, ∞), but the flow blows up at t = 1 and the displayed estimate contains factors (1 − t)^δ; the intended time interval is presumably [0, 1).","section":"§4, Theorem 4.2"},{"comment":"The notation ∇_{W_i} and the action of D_W on 1-forms should be defined before Eq. (2.4), since the second displayed expression for L uses this notation without prior explanation.","section":"§2.1, Eq. (2.4)"},{"comment":"The formula for u[a]_j contains the fragment 'u_j(τ)dσ' for j = 1, ..., I; this should be completed, presumably with an integral or with the coefficient a_j.","section":"§3.2, end of proof of Lemma 3.1"},{"comment":"The symbols φ and φ are used inconsistently in the displayed formulas for A; please unify the notation.","section":"§2.2, Eq. (2.5)"}],"recommendation":"major_revision","confidential_remarks":"I recommend major revision rather than rejection because the overall stable-manifold scheme is recognizable and the obstacle is a localized but load-bearing algebraic/spectral gap. The authors should be asked to provide a complete derivation of (2.4)–(2.5), to identify the intended linearized operator unambiguously, and to prove Lemma 2.1 for that operator. If the corrected operator turns out not to be self-adjoint with discrete spectrum, the spectral framework would need substantial reworking, so this point should be verified before further reliance on the fixed-point machinery."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a serious stable-manifold construction, and the non-equivariant setting is genuinely new. But the paper has a load-bearing problem in Section 2.2 that needs fixing before the main theorem is established.\n\nWhat is new: prior work (Donninger–Schörkhuber, Glogić–Schörkhuber) only handled SO(n)-equivariant stability and blowup. Here the authors build an infinite-dimensional family of solutions of the rescaled de-Turck flow converging to Weinkove's soliton W, without any symmetry assumption, for 5 ≤ n ≤ 9, and then use a cutoff perturbation to produce a solution not gauge-equivalent to any equivariant one. That is a real step beyond the literature. The refined Kato inequality in Lemma 3.2 and the weighted parabolic Hölder setup are sensible tools; the nonlinear fixed-point argument in Section 4 follows the standard blueprint and, conditional on the linear theory, looks coherent. The citation pattern is honest: W and the spectral background come from Weinkove and DS/GS, and the paper flags the missing non-equivariant gap.\n\nThe soft spot is real and central. In Eq (2.5) the two displayed formulas for A = -U^{-1}LU are not the same operator. Conjugating the term 2[W_i, ∂_i u_j] by U = e^{|·|^2/8} produces a y_i[W_i, φ_j] term that appears in neither display. The coefficient of [φ_i, F_Wij] changes from 2 in the first display to 1 in the second. The zeroth-order potential in the first display is 1/16(4(n-2)-|y|^2) - 1/2, while the second display has 1/16(|y|^2 - 4(n-2)); these differ by more than sign. Moreover, (D_W D_W^* + D_W^* D_W) contributes -2Δ plus a curvature term, so it cannot equal the first display's -Δ plus 2[W_i, ∂_i φ_j]. One of these expressions is a transcription error, or the operator L as written is not the one being analyzed.\n\nThis matters because Lemma 2.1 is load-bearing: it supplies the discrete L^2_ρ eigenbasis, the finite non-positive spectrum, and the projector Π_{>0} that defines the initial condition and the explicit solution formula (3.4). The proof of Lemma 2.1 runs on the second display and on self-adjointness of A. If the true conjugated operator has the extra derivative and potential terms, then the spectral decomposition used everywhere is not established. The rest of the paper cannot be certified until this is repaired.\n\nWho this is for: people working on Yang–Mills flow singularity formation and stable-manifold constructions near solitons. The paper deserves a serious referee despite my skepticism about the current state of (2.5), because the result, if fixable, is significant and the surrounding machinery is substantive. I would not cite it or put it in a reading group until the spectral lemma is corrected.","headline":"Non-equivariant Type-I blowups are the right target and the strategy is standard, but the conjugated operator in (2.5) is written inconsistently and the spectral lemma built on it is load-bearing; needs a fix before the main theorem can be trusted.","tokens_in":27105,"tokens_out":7613,"would_cite":false,"duration_ms":62214,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C44","58E15","35B44"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs an infinite-dimensional family of Yang-Mills flow solutions that converge to the equivariant shrinking soliton up to gauge, without imposing any symmetry on the perturbation.","keywords":["Yang-Mills flow","Type-I blowup","homothetically shrinking soliton","de-Turck flow","spectral stability","non-equivariant solutions","Gaussian weighted spaces","Kato inequality"],"falsifier":"Compute the spectrum of the linearized operator $-L$ in the Gaussian $L^2$ space: if any essential spectrum reaches or crosses zero, or if the eigenfunctions fail to form a complete orthonormal basis, then the projection formula (3.4) and all subsequent contraction estimates collapse. A more targeted test: exhibit an eigenfunction with eigenvalue $\\lambda > 1/2$ whose pointwise growth exceeds $C r^{2\\lambda - 1}$; Lemma 3.10 asserts no such eigenfunction exists, and that growth bound is used to control the initial data and the non-equivariant solution.","tokens_in":26030,"feed_emoji":"💥","tokens_out":8184,"duration_ms":64310,"temperature":0.7,"pith_summary":"The paper proves that the Yang-Mills flow on $\\mathbb{R}^n$ with structure group $SO(n)$, for $5 \\leq n \\leq 9$, admits an infinite-dimensional family of solutions that blow up at a finite time and converge, modulo gauge transformations, to the known $SO(n)$-equivariant homothetically shrinking soliton $W$. This is the first construction of such converging solutions without imposing $SO(n)$-equivariance on the perturbation, so stability of $W$ holds in the full space of connections. The family is parameterized by arbitrary small initial connections lying in an infinite-dimensional weighted Hölder space, subject only to a spectral projection condition that removes the finitely many non-positive modes. As a corollary, the paper constructs a non-equivariant, asymmetric Type-I blowup solution. Consequently, the stable manifold of $W$ in the gauge-fixed flow is infinite-dimensional.","feed_headline":"Asymmetric blowups found for Yang-Mills flow in dimensions 5-9","feed_subtitle":"Every small perturbation of the known soliton yields a converging solution, no symmetry assumed.","key_machinery":"The load-bearing object is the spectral projection $\\Pi_{>0}$ of the linearized operator $-L$ of the rescaled Yang-Mills de-Turck flow at the soliton $W$, acting on the Gaussian-weighted space $L^2_\\rho$ with $\\rho = e^{-|x|^2/4}$. Lemma 2.1 gives $-L$ a complete orthonormal eigenbasis with finitely many non-positive eigenvalues, so $\\Pi_{>0}$ selects the modes that decay exponentially in the similarity time $\\tau$. The explicit solution formula (3.4) represents the linearized solution using forward-in-time integrals for positive eigenvalues and backward integrals for non-positive ones, and a refined Kato inequality for the Frobenius norm (Lemma 3.2) supplies global sup-norm control that standard comparison arguments cannot give for tensor-valued sections. These pieces combine with weighted parabolic Hölder spaces to run the contraction-mapping argument that produces the nonlinear solution.","core_discovery":"Fix $5 \\leq n \\leq 9$ and let $W$ be the explicit homothetically shrinking soliton found by Weinkove. For every smooth connection $\\tilde A_0$ sufficiently close to $W$ in a weighted Hölder norm, there is a smooth solution $A$ of the Yang-Mills flow, defined up to time $1$, such that the spectral projection condition $\\Pi_{>0}(A(\\cdot,0)-W-\\tilde A_0)=0$ holds and, after applying a time-dependent gauge transformation, $A$ converges to $W$ at the rate $(1-t)^\\delta$ relative to the soliton's $L^\\infty$ scale. The proof works with the rescaled Yang-Mills de-Turck flow, whose linearization at $W$ is a self-adjoint operator with discrete spectrum in a Gaussian $L^2_\\rho$ space; projecting initial data onto the positive spectral subspace and running a fixed-point argument yields the solution. This establishes, in the paper's language, that the unstable and neutral directions can be freely chosen while the positive modes control convergence, and it yields the corollary that a non-$SO(n)$-equivariant solution exists.","pith_inferences":["Editorial inference: the explicit translational eigenfunctions $F_{W\\alpha}$ with eigenvalue $-1/2$ should generate a family of blowup solutions with drifting centers; tracking the coefficients of these modes would quantify the drift rate.","Editorial inference: because the lowest eigenfunction (eigenvalue $-1$) and the translational modes (eigenvalue $-1/2$) control the leading decay, generic non-equivariant solutions should approach $W$ with a small spatial translation; this is a testable prediction from the coefficients of $A(\\cdot,0)-W$ on those two eigenspaces.","Editorial inference: the method is not tied to the specific structure group; it should adapt to any gauge-gradient flow whose linearized operator at a soliton has discrete Gaussian spectrum with finitely many non-positive eigenvalues, such as harmonic map flow."],"forward_implications":["There exist asymmetric Type-I blowup solutions of the Yang-Mills flow, i.e., solutions whose curvature blows up with the maximal self-similar rate, that are not gauge equivalent to any $SO(n)$-equivariant solution.","The stable manifold of $W$ in the full non-equivariant, gauge-fixed Yang-Mills flow is infinite-dimensional: the admissible perturbations $\\tilde A_0$ range over an infinite-dimensional weighted Hölder space.","In dimensions $5$ through $9$, the equivariant soliton $W$ is dynamically stable in the full space of connections, not just in the $SO(n)$-equivariant class, in the sense that every sufficiently small perturbation admits a smooth solution converging to $W$ up to gauge.","The convergence is exponential in similarity coordinates, giving the explicit bound $\\frac{\\|W(\\cdot,t)-S(t)^* A(\\cdot,t)\\|_{L^\\infty}}{\\|W(\\cdot,t)\\|_{L^\\infty}} \\leq M\\|\\tilde A_0\\| (1-t)^\\delta$ for all $t \\in [0,\\infty)$.","The proof supplies a concrete non-symmetric initial datum by cutting off $W$ far from the origin, so the asymmetry of the resulting blowup solution appears at arbitrarily large spatial scales."],"supporting_citations":[{"why":"Supplies the explicit homothetically shrinking soliton $W$ and its curvature form, the equilibrium around which the whole construction is built.","marker":"[Wei04]"},{"why":"Establishes equivariant stability for $n=5$ and identifies the lowest eigenfunction, the symmetric-case precedent the paper extends.","marker":"[DS19]"},{"why":"Extends equivariant stability to $5 \\leq n \\leq 9$, the exact dimensional range of the main theorem.","marker":"[GS20]"},{"why":"Introduces the Yang-Mills de-Turck flow formulation used to turn the flow into an elliptic linear problem with gauge covariance.","marker":"[Str94]"},{"why":"Provides the spectral-theoretic compactness criterion invoked in Lemma 2.1 to prove discreteness and completeness of the eigenbasis.","marker":"[RS78]"},{"why":"Gives growth bounds for the lowest eigenfunction that Lemma 3.9 and Lemma 3.10 refine into pointwise bounds for all eigenfunctions.","marker":"[BW17]"},{"why":"Supplies the barrier-function and rescaling technique that Lemma 3.3 adapts, via Kato's inequality, to tensor-valued sections.","marker":"[CCMS24]"},{"why":"Provides the Gaussian Sobolev inequality used in the eigenfunction growth estimate that controls the non-positive modes.","marker":"[Eck00]"},{"why":"Provides the non-standard parabolic Schauder estimate used in Lemma 3.10 to obtain Hölder regularity of eigenfunctions in exterior regions.","marker":"[Kne80]"}],"fun_headline_variants":["Infinite family of asymmetric blowups for Yang-Mills flow","Asymmetric Type-I blowups for Yang-Mills flow in dimensions 5-9","Infinite-dimensional family of Yang-Mills blowup solutions","Asymmetric blowups for Yang-Mills flow found in dimensions 5-9"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction assumes that near the soliton every perturbation splits cleanly into a finite number of decaying modes and an infinite number of growing modes, with no continuous spectrum in between; if that spectral picture failed, the projection that selects converging initial data would not exist.","fun_headline_variants_meta":{"raw":{"variants":["Infinite family of asymmetric blowups for Yang-Mills flow","Asymmetric Type-I blowups for Yang-Mills flow in dimensions 5-9","Infinite-dimensional family of Yang-Mills blowup solutions","Asymmetric blowups for Yang-Mills flow found in dimensions 5-9"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000981,"raw_usage":{"total_tokens":4115,"prompt_tokens":848,"completion_tokens":3267,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":464,"completion_tokens_details":{"reasoning_tokens":3188}},"tokens_in":464,"tokens_out":3267,"duration_ms":21166,"temperature":1.0,"reasoning_tokens":3188,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:21:24.297097+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the spectrum of the linearized operator $-L$ in the Gaussian $L^2$ space: if any essential spectrum reaches or crosses zero, or if the eigenfunctions fail to form a complete orthonormal basis, then the projection formula (3.4) and all subsequent contraction estimates collapse. A more targeted test: exhibit an eigenfunction with eigenvalue $\\lambda > 1/2$ whose pointwise growth exceeds $C r^{2\\lambda - 1}$; Lemma 3.10 asserts no such eigenfunction exists, and that growth bound is used to control the initial data and the non-equivariant solution.","supporting_citations":[],"review_version":1}