{"id":"9080f3fb-ae76-43cd-bfda-1359371857b4","arxiv_id":"1908.08274","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Invading parallel stripe fronts in the planar Swift-Hohenberg equation select a unique speed and wavenumber, while retreating fronts form one-parameter families; almost planar fronts bifurcate from parallel fronts near the snaking region.","lead":"This paper develops a numerical method to track moving stripe patterns that invade or retreat in the Swift-Hohenberg equation. It shows how the speed and stripe spacing of invading fronts are selected, and how two-dimensional fronts bifurcate near the region where patterns are pinned.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 1.1's uniqueness conclusion rests on an unproved and internally inconsistent transversality hypothesis (Hypothesis 3); the numerics certify discretization error, not the absence of extra neutral modes.","rationale":"The paper's central theoretical claim is the conditional Proposition 1.1, and its proof in Appendix A.1 hinges on Hypothesis 3. The reader identified this as the weakest assumption, and I agree: without a two-dimensional translation kernel and algebraic simplicity of the relevant zero eigenvalue, the Lyapunov-Schmidt reduction does not yield uniqueness of (k_x, omega) for invading fronts. The numerical convergence checks in Figures 12 and 13 are genuine evidence that a particular computed branch is well resolved, but they do not probe the spectral hypothesis. I also note that Hypothesis 3 as written is formally inconsistent because a two-dimensional kernel cannot be algebraically simple; the intended condition must be stated separately for L^2_alpha and L^2_{-alpha}. This strengthens rather than replaces the reader's concern. I do not recommend changing the verdict to rejection: the theorem is explicitly labeled as conditional, the paper is primarily a numerical bifurcation study, and the concrete spectral test is feasible and would settle whether the assumption holds. The appropriate outcome remains conditional acceptance, which is what the reader already recommended.","tokens_in":56792,"tokens_out":12337,"duration_ms":141055,"concrete_test":"Using a converged parallel invasion front at fixed (mu, nu) = (0, 1.6) from Section 7.1, discretize the conjugated linearized operator on the exponentially weighted space, e^{-alpha|rho|} L e^{alpha|rho|} with L = omega(d_rho - d_tau) - (1 + k_x^2 d_rho^2)^2 - mu + f'(u*), using the same pseudo-spectral tau and fourth-order finite-difference rho discretization, with alpha > 0 small, domains L_rho = 40 pi and 80 pi, and resolutions N_rho = 400, 800 and N_tau = 20, 40. Compute the ten smallest-magnitude eigenvalues. Hypothesis 3 predicts exactly one zero eigenvalue in L^2_alpha (the localized mode (d_rho - d_tau)u*) with algebraic multiplicity 1, exactly two zero eigenvalues in L^2_{-alpha} (d_rho u* and d_tau u*), and all other eigenvalues with modulus bounded below uniformly in L_rho.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Proposition 1.1 and the interpretation of the numerical branches in Section 7 depend on Hypothesis 3 in Appendix A.1: for the linearization L about a parallel invasion front in L^2_{-alpha}, the kernel is exactly spanned by d_rho u* and d_tau u*, and the zero eigenvalue is algebraically simple. This is a transversality condition that is not proved for the Swift-Hohenberg equation, and the convergence studies in Section 6.4 (Figures 12-13) only certify discretization error of one computed solution, not the absence of additional neutral modes. The condition as written is also internally inconsistent: a two-dimensional kernel cannot have an algebraically simple zero eigenvalue; what the proof actually needs is that the point-spectrum zero eigenvalue in L^2_alpha is simple and that the parameter derivatives (d_kx, d_psi, d_omega) F_v cover the remaining cokernel directions of DF_v. If an additional neutral mode exists, for example a stripe sideband mode becoming resonant with the front, the Lyapunov-Schmidt reduction in Proposition A.1 collapses: there may be folds where (k_x, omega) cease to be locally unique, or a continuum of invading fronts not captured by the far-field core BVP. The numerical branch would then be evidence of a particular solution branch, not of pattern selection. This is a correctness risk, not a disagreement with consensus.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a numerical continuation framework for depinning (invasion) fronts in the planar Swift-Hohenberg equation in the bistable region outside homoclinic snaking. The main theoretical result is a conditional selection principle (Proposition 1.1): transversely invading parallel fronts are claimed to select a unique speed and far-field wavenumber, whereas retreating fronts are claimed to form one-parameter families. The numerical methodology adapts the far-field core decomposition of Lloyd & Scheel to moving fronts, adding phase conditions for the transition frequency and far-field wavenumber. Extensive computations are reported for parallel, oblique, perpendicular, and almost-planar fronts in the quadratic-cubic and cubic-quintic SH equations, including two-parameter sweeps, comparison to semi-analytical theory near the snaking edge, and application to the invasion of localized stripe patches.","tokens_in":57069,"tokens_out":6432,"duration_ms":64911,"significance":"If the claims are correct, the paper makes a valuable methodological contribution: it provides a concrete, systematic way to compute pattern-selecting invasion fronts in a class of pattern-forming PDEs, going beyond time-stepping and beyond the weakly nonlinear regime. The far-field core decomposition with phase conditions is a natural and useful extension of prior work, and the convergence studies for the 1D parallel-front solver (Figures 12-13) are careful and informative. The comparison with the semi-analytical |δ|^{1/2} scaling law (Figure 15) gives a meaningful benchmark, and the paper is honest about several open issues, including the lack of a proof of the needed transversality. However, the central uniqueness theorem rests on an unproved and, as stated, internally inconsistent transversality hypothesis, and the main numerical examples for the quadratic-cubic equation are computed in a regime that appears to violate the theorem's stability hypothesis. The almost-planar front coexistence, one of the headline observations, is supported only by computations at a single transverse resolution without a published convergence check.","major_comments":[{"comment":"Hypothesis 3 as stated is internally inconsistent: it says the kernel of L in L^2_{-α} is two-dimensional, spanned by ∂_ρ u* and ∂_τ u*, and then says the eigenvalue λ=0 is algebraically simple. A two-dimensional kernel in that space precludes algebraic simplicity of the zero eigenvalue. The proof of Proposition A.1 actually needs a different statement: in the exponentially weighted space L^2_α the kernel should be one-dimensional (e.g., spanned by an exponentially localized combination of ∂_ρ u* and ∂_τ u*) and the parameter derivatives (∂_{kx}, ∂_ψ, ∂_ω) F_v should cover the remaining cokernel directions of D F_v. The hypothesis as written is therefore not merely unverified but ill-posed, and it is load-bearing for the uniqueness part of Proposition 1.1. The convergence tests in Section 6.4 (Figures 12-13) certify discretization error of a single computed solution; they do not certify the absence of additional neutral modes. If an extra mode, such as a resonant stripe sideband, exists, the Lyapunov-Schmidt reduction of Proposition A.1 fails and the branches computed in Section 7 could be folds rather than genuine selection curves.","section":"Appendix A.1, Hypothesis 3"},{"comment":"The uniqueness/selection theorem assumes temporal stability of the far-field stripe solution, but the main numerical study of parallel fronts in Section 7.1 uses the quadratic-cubic SH equation at ν=1.6, for which the paper itself states (Section 2) that stripes are typically unstable in the bistable region to hexagonal perturbations. No argument is given to show that the hexagon instability is irrelevant for the 1D front problem, nor is the stability hypothesis verified for the computed branches. Consequently, the interpretation of Figure 14 as a confirmation of Proposition 1.1 is not supported by the stated hypotheses of that proposition. The paper needs to either restrict the theoretical claim to stable regimes, prove that the relevant stability notion is only the longitudinal one, or explicitly mark the quadratic-cubic results as being outside the theorem's assumptions.","section":"Proposition 1.1 and Section 7.1"},{"comment":"The claim that almost planar invasion fronts bifurcate off parallel fronts and coexist near the snaking region is a headline observation, but it is based on computations with N_γ = N_τ = 16 and no convergence study in the transverse direction. The parallel branch itself is reported to be zig-zag unstable for the whole bistable region (Figure 19, left panel), which makes it unclear whether the observed bifurcation is a genuine solution bifurcation or a numerical artifact from an under-resolved transverse discretization. The paper should provide a mesh-refinement study in γ and τ for the almost-planar branches, and it should report the extrapolated values of ω and k_x with error estimates at the reported fold location, before this claim can be considered quantitatively supported.","section":"Section 7.3, Figure 19"}],"minor_comments":[{"comment":"The text says 'invading fronts with ω<0' in Lemma 1 and in the statement of Proposition A.1, but throughout Section 6 and the introduction ω is defined as a positive quantity (ω = c k_x, c>0 for invading fronts). Please clarify the sign convention.","section":"Lemma 1 and Proposition A.1"},{"comment":"The manuscript says there are 'three types of possible depinning stripe fronts: parallel, oblique, perpendicular and almost-planar stripe fronts' but then lists four types; please correct the count or the list.","section":"Section 1, paragraph after Figure 2"},{"comment":"The symbol L is used both for the linearized operator and for the domain length L_ρ; also the definition of L in (6.9) uses a minus sign before the biharmonic term while the operator in (6.7a) has a plus sign before it; please harmonize the notation.","section":"Section 6.4, equation (6.9)"},{"comment":"The fitted scaling law is reported as ω = 1.18 |μ-μ_1|^{0.5292} with no error bars or fit-quality measures; given that the exponent is expected to be 0.5, a confidence interval on the fitted exponent should be stated, along with the number of data points used and the residuals.","section":"Figure 15"},{"comment":"The text states that 'IVP suggests stability for smaller ky values' but no supporting IVP results or stability computation is shown; please provide evidence or rephrase as a conjecture.","section":"Section 7.3, last paragraph"},{"comment":"Several references are listed as 'Submitted 2018' or 'Submitted 2019'; if any have since appeared, please update them; otherwise state the status explicitly in the bibliography.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper contains a promising numerical framework and a substantial body of computations, but the central theoretical claim is presented as a theorem under a hypothesis that is both unverified and, as written, internally inconsistent. This is likely fixable: the authors should restate the transversality hypothesis in the form actually used in the Lyapunov-Schmidt proof, clearly separate the condition from its verification, and either prove it for the SH equation or explicitly label it as an open conjecture. The mismatch between the stability hypothesis of Proposition 1.1 and the numerically studied regimes (especially the quadratic-cubic case) should also be addressed. If the authors can revise the theory to be internally consistent and add convergence evidence for the 2D almost-planar computations, the paper could be acceptable; in its current form, however, the main claims outrun the evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a serious numerical study of depinning fronts in the two-dimensional Swift–Hohenberg equation, and the main results hold up. I would send it to referees without hesitation.\n\nWhat is new: the far-field core decomposition is extended from stationary grain boundaries to moving invasion fronts, which yields a BVP that can be continued for the front speed and the far-field wavenumber. That solves a problem the paper correctly identifies as open — nobody had been able to do numerical bifurcation analysis of these depinning fronts. The two headline observations, the dip in the selected wavenumber away from the snaking region and the bifurcation from parallel to almost-planar invasion fronts, are genuinely new and documented carefully. The comparison with the semi-analytical |δ|^(1/2) scaling is honest, and the weakly nonlinear comparison acts as a legitimate external benchmark, not a fitted result. The lineage from the author's own grain-boundary work is clear and acknowledged; that is self-reference, not circularity.\n\nThe soft spot worth naming: the analytical framework rests on Hypothesis 3 (Appendix A.1), a transversality condition not proved for SH. On the stress-test note you passed along: the charge of internal inconsistency is overstated. The wording of Hypothesis 3 is sloppy, but Lemma 2 clarifies the standard picture — the kernel in the unweighted space is two-dimensional from the two translations, while in the weighted space L^2_α only the localized combination survives, so the point-spectrum eigenvalue is simple. The proof goes through the weighted space, so the apparent contradiction is verbal, not load-bearing. What remains genuinely unverified is the transversality itself; the convergence studies in §6.4 certify discretization error, not the absence of extra neutral modes. That is a real caveat, and an extra neutral mode would break the selection conclusion. But the proposition is labeled conditional and the numerics are presented as numerical results, which is the right framing.\n\nMinor complaints: no code or data shipped, which matters more for a numerical-methods paper; and most bifurcation diagrams carry no error bars, which is common in this community and not a blocker. The author is also candid about the limits of the semi-analytical perturbation theory in §5, which I counted in the paper's favor.\n\nWho this is for: people working on localized patterns, depinning, and numerical continuation of fronts. I would cite it in the next year if I were working on those problems.\n\nRecommendation: accept after revision, conditional on tightening the presentation of Hypothesis 3 and making the numerical details more accessible.","headline":"A serious, publishable numerical study of depinning fronts with genuinely new results; the numerics are careful and the theory is honestly conditional on an unverified transversality assumption.","tokens_in":57586,"tokens_out":6287,"would_cite":true,"duration_ms":55705,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B36","35B32","35C07","65P30"],"pacs":[],"model":"deepseek-v4-flash","headline":"Invading stripe fronts in the Swift-Hohenberg equation select a unique speed and wavenumber, while retreating fronts form one-parameter families.","keywords":["Swift-Hohenberg equation","depinning fronts","homoclinic snaking","pattern selection","travelling waves","numerical continuation","far-field core decomposition","stripe patterns"],"falsifier":"Compute, at a numerically converged invasion front for representative parameters in the bistable region, the eigenvalues and eigenfunctions of the linearised operator $L$ in the exponentially weighted space $L^2_{-\\alpha}$; an extra near-zero eigenvalue, or a zero eigenvalue of algebraic multiplicity greater than one, would refute Hypothesis 3 and with it the uniqueness part of Proposition 1.1. A second, independent test is to run two time simulations starting from fronts with different initial stripe wavenumbers or phases and check whether both converge to the same $c$ and $k_x$.","tokens_in":1869,"feed_emoji":"🌊","tokens_out":2772,"duration_ms":84675,"temperature":0.7,"pith_summary":"The paper addresses the motion of stripe patterns into a quiescent state in the two-dimensional Swift-Hohenberg equation, focusing on the parameter region just outside the homoclinic snaking window. Its central claim, Proposition 1.1, is that an invading parallel front selects a unique invasion speed $c$ and far-field wavenumber $k_x$, whereas a retreating parallel front comes in a one-parameter family with speed $c=c(k_x)$. To make this principle usable, the paper builds a boundary-value problem from a far-field core decomposition and traces the fronts by numerical continuation. The numerics find that almost-planar invasion fronts in the cubic-quintic equation bifurcate off parallel fronts near the snaking region and do not exist far away, and that parallel fronts may regain transverse stability at high speed. If the claims are right, front propagation outside the snaking window is a structured pattern-selection phenomenon that can be mapped systematically.","feed_headline":"Invading fronts pick one speed and wavelength; retreating fronts don't","feed_subtitle":"Depinning stripe fronts in the Swift-Hohenberg equation obey a selection rule that numerical continuation now maps in detail.","key_machinery":"The load-bearing object is the far-field core decomposition, which writes a front as the product of a far-field stripe $u_s(k_x(\\rho\\pm\\tau+\\psi);k_x)$ with a cut-off $\\chi(\\rho)$ plus a remainder $v(\\rho,\\tau;\\omega)$ that is localised near the interface. Posed in exponentially weighted spaces $L^2_\\alpha$, this makes the linearised operator Fredholm so that Newton and continuation methods apply, with two integral phase conditions fixing the translational mode and the far-field wavenumber; the unknowns solved for are the remainder $v$, the wavenumber $k_x$, and the transition frequency $\\omega=c k_x$. The proof of Proposition 1.1 uses a Lyapunov-Schmidt reduction in this setup, and the stationary selection theory enters through the conserved quantities $H(u)$ and $S(u)$ that force a stationary front's far-field stripes to sit on the zero-Hamiltonian set.","core_discovery":"The paper's main result is a pattern-selection dichotomy for parallel depinning fronts. Assuming the far-field stripe is temporally stable and the front is transverse, an invading front survives small perturbations and selects a unique invasion speed $c$ and far-field wavenumber $k_x$ for fixed $\\mu$ and $\\nu$; a retreating front instead forms a one-parameter family parameterised by $k_x$, with $c=c(k_x)$. The numerical companion result, obtained by solving the far-field core decomposition boundary-value problem, is that almost-planar invasion fronts in the cubic-quintic Swift-Hohenberg equation (nonlinearity $f(u)=\\nu u^3-u^5$) bifurcate from parallel invasion fronts and coexist close to the homoclinic snaking region, while far away no almost-planar fronts exist and parallel fronts can regain transverse stability once they propagate above a critical speed.","pith_inferences":["If the pattern-selection dichotomy holds beyond the Swift-Hohenberg equation, then bistable pattern-forming systems generally have a computable invasion speed and wavelength, so patch morphology in reaction-diffusion or convection models could be predicted from front data alone.","Hypothesis 3 could be certified numerically by solving the adjoint eigenvalue problem at the computed fronts; a numerical proof of a simple algebraically isolated zero eigenvalue would upgrade Proposition 1.1 from assumed to verified for the computed branches.","The wavenumber dip near the snaking edge is consistent with exponentially small splitting between stable and unstable manifolds, suggesting that matched-asymptotics methods could predict the location and depth of the dip as a function of $\\mu$ and $\\nu$.","The same boundary-value construction should apply to fronts whose far field is a different cellular pattern, such as hexagons, and to systems with nonlocal or discrete spatial structures, where the essential-spectrum obstruction is analogous."],"forward_implications":["Invading parallel depinning fronts can be path-followed in $(c,k_x,\\mu,\\nu)$-space, so folds and stability changes are located by continuation rather than inferred from time simulations.","The selected wavenumber begins at the Hamiltonian-selected value at the snaking edge, dips, then increases; this non-monotonic dip is a quantitative target for exponential-asymptotic theories of pinning.","Almost-planar invasion fronts exist only close to the snaking region, where they coexist with parallel fronts and travel slower; far away, parallel fronts are the only invasion mode and can become transversely stable at high speed.","Patch growth on the plane is governed by the parallel and perpendicular front speeds: the parallel front fixes the stepping rate in one direction while the perpendicular front sets the other, matching the observed bulging of growing stripe patches."],"supporting_citations":[{"why":"Supplies the far-field core decomposition and continuation method for planar grain boundaries, which the paper adapts to moving depinning fronts.","marker":"[37]"},{"why":"Provides the Fredholm-index and pattern-forming front framework for parallel and oblique stripes in Swift-Hohenberg systems, used for front persistence and index computations.","marker":"[23]"},{"why":"Supplies the semi-analytical perturbation theory for transition time near the snaking folds, used for comparing the square-root scaling of the invasion speed.","marker":"[10]"},{"why":"Establishes the existence and stability of stationary almost-planar fronts and worm patches in the planar Swift-Hohenberg equation, forming the basis for the bifurcation and patch simulations.","marker":"[4]"},{"why":"Provides the weakly nonlinear travelling-front solutions and selection predictions for the amplitude equation, used as a comparison for wavenumber and speed selection.","marker":"[50]"},{"why":"Supplies the first conserved quantity and stationary front selection results in two dimensions, underlying the Hamiltonian selection discussion.","marker":"[36]"},{"why":"Provide the weighted-space Fredholm theory and relative Morse index calculations for modulated fronts, which underpin the proof of Proposition 1.1.","marker":"[53,54]"}],"fun_headline_variants":["Invading fronts select unique speed; retreating fronts don't","Stripe fronts invade at one speed, retreat at many","Almost-planar fronts appear only close to snaking","Parallel fronts regain stability above critical speed"],"cache_read_input_tokens":59648,"weakest_assumption_plain":"The result assumes a transversality condition (Hypothesis 3): the kernel of the linearisation about the front in $L^2_{-\\alpha}$ is exactly two-dimensional, spanned by spatial and temporal translations, and the zero eigenvalue is algebraically simple; the paper states this without proving it for the Swift-Hohenberg equation.","fun_headline_variants_meta":{"raw":{"variants":["Invading fronts select unique speed; retreating fronts don't","Stripe fronts invade at one speed, retreat at many","Almost-planar fronts appear only close to snaking","Parallel fronts regain stability above critical speed"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000229,"raw_usage":{"total_tokens":1527,"prompt_tokens":1041,"completion_tokens":486,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":657,"completion_tokens_details":{"reasoning_tokens":423}},"tokens_in":657,"tokens_out":486,"duration_ms":5142,"temperature":1.0,"reasoning_tokens":423,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:44:27.429939+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, at a numerically converged invasion front for representative parameters in the bistable region, the eigenvalues and eigenfunctions of the linearised operator $L$ in the exponentially weighted space $L^2_{-\\alpha}$; an extra near-zero eigenvalue, or a zero eigenvalue of algebraic multiplicity greater than one, would refute Hypothesis 3 and with it the uniqueness part of Proposition 1.1. A second, independent test is to run two time simulations starting from fronts with different initial stripe wavenumbers or phases and check whether both converge to the same $c$ and $k_x$.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the far-field core decomposition and continuation method for planar grain boundaries, which the paper adapts to moving depinning fronts."},{"cited_title":"Goh and A","cited_arxiv_id":null,"evidence_quote":"Provides the Fredholm-index and pattern-forming front framework for parallel and oblique stripes in Swift-Hohenberg systems, used for front persistence and index computations."},{"cited_title":"Burke and E","cited_arxiv_id":null,"evidence_quote":"Supplies the semi-analytical perturbation theory for transition time near the snaking folds, used for comparing the square-root scaling of the invasion speed."},{"cited_title":"Avitabile, D","cited_arxiv_id":null,"evidence_quote":"Establishes the existence and stability of stationary almost-planar fronts and worm patches in the planar Swift-Hohenberg equation, forming the basis for the bifurcation and patch simulations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the weakly nonlinear travelling-front solutions and selection predictions for the amplitude equation, used as a comparison for wavenumber and speed selection."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the first conserved quantity and stationary front selection results in two dimensions, underlying the Hamiltonian selection discussion."}],"review_version":1}