{"id":"73f29a38-60a5-4647-998d-db2e95e143ca","arxiv_id":"2411.13094","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Stationary shock profiles of the Lax-Wendroff scheme are nonlinearly orbitally stable under a spectral stability assumption, with explicit weighted decay rates, despite the scheme's known dispersive ell-1 instability.","lead":"This paper proves that the shock profiles of the Lax-Wendroff scheme, a standard second-order method for conservation laws, are nonlinearly orbitally stable for convex or concave fluxes, with explicit decay rates. It provides a quantitative picture of the Green's function, including the dispersive oscillations ahead of the shock, and a template for stability proofs of other high-order schemes.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2.5's stated parameter range is incompatible with the paper's own linear estimates: for β<1/3 the claimed ℓ∞ decay n^{-σ-11/24} is faster than the semigroup bound n^{-σ-β-1/8}, so the proof of the ℓ∞ estimate fails for admissible parameters.","rationale":"The reader's chosen weakest assumption, Assumption 1, is an honest external caveat that the authors themselves state and discuss; the paper even proves Assumption 1 for convex/concave fluxes. It does not undermine the main theorem under its stated hypotheses. The truly load-bearing concern is internal: the proof of Theorem 2.5 uses decay rates that are faster than those provided by the paper's own linear theory, Theorem 4.3. The ℓ∞β rate σ+11/24 cannot follow from (4.37b) when β<1/3 because the linear semigroup is only shown to decay like n^{-σ-β-1/8}. Since the theorem explicitly allows such β (e.g., β=0.2, σ=0.23), the induction step contains a genuine algebraic error, not just a missing assumption about the flux. The same error recurs in the treatment of the nonlinear terms via (4.38b)-(4.38c), where for β<1/4 the exponents are replaced by larger values, invalidating the summation estimates. Consequently, the central claim as stated is not established; the paper could be repaired by restricting the theorem to β≥1/3 (and adjusting the nonlinear summation arguments) or by proving sharper linear estimates. This warrants conditional acceptance rather than acceptance in the current form. I disagree with the reader's attribution of the weakest point to Assumption 1, because the spectral assumption is explicitly flagged and verified for the main convex/concave cases, whereas the exponent issue affects the theorem even when Assumption 1 holds.","tokens_in":86279,"tokens_out":25188,"duration_ms":210389,"concrete_test":"Check the exponent in the first inequality of the ℓ∞β estimate in §5.2.2: substitute γ=σ+β+1/8 into (4.37b)'s exponent γ−β+min(1/3,β) and compare with σ+11/24. For β=0.2, σ=0.23, γ−β+min(1/3,β)=0.555<0.688, so the proof's bound on L^{n+1}p0 is invalid. Additionally, inspect the nonlinear sums: for β=0.2, σ=0.23, verify whether Lemma 5.3's hypotheses hold with the correct exponents a=2β+1/3 and a=2β−1/8 rather than the used β+7/12 and β+1/8; they fail (2β+σ<2/3).","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Even under Assumption 1, the proof of the central nonlinear orbital stability theorem (Theorem 2.5) contains an exponent mismatch. Theorem 2.5 asserts, for any β,σ≥0 with β+σ≥5/12 and 0≤σ<β+1/8, the bound ∥p^n∥ℓ∞β ≤ C0(1+n)^{-σ-11/24}∥h∥ℓ1γ, where γ=σ+β+1/8. In the induction step (§5.2.2, first term of the ℓ∞β estimate), the linear evolution L^{n+1}p0 is bounded using Theorem 4.3(4.37b), whose exponent is γ2−γ1+min(1/3,γ1) = γ−β+min(1/3,β) = σ+1/8+min(1/3,β). For β<1/3 this equals σ+β+1/8, which is strictly smaller than σ+11/24; hence the claimed inequality ∥L^{n+1}p0∥ℓ∞β ≤ C(2+n)^{-σ-11/24}∥p0∥ℓ1γ is unjustified. Example: β=0.2, σ=0.23 satisfies all hypotheses (β+σ=0.43≥5/12, σ<β+1/8=0.325), but the semigroup estimate gives decay n^{-0.555}, slower than the asserted n^{-0.688}. The same problem propagates to the nonlinear sums: for β<1/4, the exponents in (4.38b) and (4.38c) are 2β+1/3 and 2β−1/8, not β+7/12 and β+1/8 as used, so Lemma 5.3 is applied outside its validity range and the claimed decay fails. Thus the main theorem is not proven on its stated parameter range; a hidden restriction β≥1/3 is required for the proof to close.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes the spectral, linear, and nonlinear stability of the piecewise constant stationary shock profile (2.7) for the Lax-Wendroff scheme (2.5)-(2.6) for scalar conservation laws. It introduces a Lopatinskii determinant (3.3) whose zero set locates the spectrum of the linearized operator L, proves for convex or concave fluxes that the only zero in the closed exterior region is the simple eigenvalue 1 (Lemma 3.2), and formulates Assumption 1 under which L has no other spectrum in the closed exterior (Theorem 2.3). The central technical content is a detailed construction of the spatial Green's function (Proposition 3.1), a decomposition of the temporal Green's function into an activation-function contribution and remainder terms governed by free constant-coefficient Green's functions (Theorem 4.1), and resulting linear decay estimates in polynomially weighted spaces (Theorem 4.3, (4.37a)-(4.38c)). A bootstrap argument in Chapter 5 uses these estimates to prove orbital stability of the family of stationary shock profiles (Theorem 2.5) for small perturbations, with rates ||p^n||_{ℓ1_β} ≤ C0(1+n)^{-σ}||h||_{ℓ1_γ} and ||p^n||_{ℓ∞_β} ≤ C0(1+n)^{-σ-11/24}||h||_{ℓ1_γ}.","tokens_in":86647,"tokens_out":9865,"duration_ms":88366,"significance":"If the parameter-range issue identified below is corrected, the paper would be a significant contribution to the stability theory of high-order numerical schemes for hyperbolic conservation laws. The explicit Lopatinskii determinant analysis for convex and concave fluxes, the fully written spatial Green's function formulas, and the sharp oscillatory integral estimates in Appendix A are concrete and checkable contributions, and the paper is honest about the failure of uniform ℓ1 estimates and about the conditional nature of Assumption 1. The conditional nonlinear stability theorem is a substantial extension of prior parabolic-type results to a dispersive setting. However, the main theorem as stated is not proven on its full parameter range because the proof uses linear and nonlinear semigroup estimates with exponents that are valid only under hidden restrictions on β. This is a load-bearing flaw that requires a corrected statement or a modified proof.","major_comments":[{"comment":"The ℓ∞_β bound for the linear term L^{n+1}p0 in the induction step is claimed as C_L(β,γ)(2+n)^{-σ-11/24}||p0||_{ℓ1_γ}. Substituting γ1=β and γ2=γ=σ+β+1/8 into Theorem 4.3(4.37b) gives the exponent γ2−γ1+min(1/3,γ1) = σ+1/8+min(1/3,β), which equals σ+11/24 only when β≥1/3. For β<1/3 the exponent is σ+β+1/8, which is strictly smaller than σ+11/24, so the claimed inequality is not a consequence of (4.37b). The theorem statement allows β<1/3, for example β=0.3 and σ=0.125 satisfy β+σ≥5/12 and σ<β+1/8, but the decay bound used in the proof fails for such parameters. The proof therefore establishes Theorem 2.5 only on the restricted range β≥1/3, or with a slower ℓ∞ rate, and the statement must be corrected accordingly.","section":"§5.2.2, Eq. (4.37b)"},{"comment":"In bounding the two nonlinear sums in the ℓ∞_β estimate, the proof replaces the factors min(1/4,β) and min(1/8,β−1/8) in (4.38b) and (4.38c) by 1/4 and 1/8, respectively, using the denominators (1+n−m)^{β+7/12} and (1+n−m)^{β+1/8}. These replacements require β≥1/4; for β<1/4 the correct exponents are 2β+1/3 and β+min(1/8,β−1/8), which are smaller than the ones used. This is a second hidden restriction in the stated parameter range. The more restrictive condition β≥1/3 imposed by the linear term subsumes it, but if the authors choose instead to weaken the ℓ∞ rate, the nonlinear part of the bootstrap would still require β≥1/4.","section":"§5.2.2, Eqs. (4.38b)-(4.38c)"}],"minor_comments":[{"comment":"The manuscript contains several typographical errors, including 'Thank's' on p. 55, 'avoir' on p. 5, 'perurbation' on p. 16, and 'encompasses' on p. 20; these should be corrected in a revision.","section":"Various"},{"comment":"The two instability examples are convincing in outline, but the phrase \"can be achieved by tuning small amplitude oscillations near the origin\" is informal; a precise construction of the flux f with the stated derivative values would make the examples fully reproducible.","section":"§3.4"},{"comment":"The notation for B_ε(0) and B_ε(0) in the definition of the region Z_{ε*,η*} is not visually distinct enough in print; the reader must carefully track which of the round or square balls is intended, so a clearer notation or an explicit reminder at each occurrence would help.","section":"§3.2, (3.21)"},{"comment":"The constants C1 and C2 in §5.1 are defined by long displays that are hard to parse; splitting them into separate displayed summands and explicitly recording the hypothesis on β used for each application of Lemma 5.3 would improve readability.","section":"§5.1"}],"recommendation":"major_revision","confidential_remarks":"This is a strong paper with substantial technical content, but the stated Theorem 2.5 is not proven on its full parameter range because the ℓ∞ semigroup estimates used in the bootstrap require β≥1/3 (and the nonlinear estimates require β≥1/4). This is fixable by restricting the hypotheses or by weakening the claimed decay rate, so I recommend major revision rather than rejection. I would also ask the authors to state the corrected range explicitly in the abstract and to comment on whether β≥1/3 is optimal for the method."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know before you read it. First, the linear theory is real work: an explicit Lopatinskii determinant, a convex/concave spectral stability proof, and a pointwise Green's function with an 'activation function' decomposition that gives sharp decay in polynomial weights without a spectral gap. That part goes beyond Smyrlis and is probably correct. Second, the main nonlinear theorem, Theorem 2.5, is not proven on its stated parameter range. The proof of the ℓ∞β estimate applies the semigroup bound (4.37b) with γ1=β, γ2=σ+β+1/8, which yields decay exponent σ+1/8+min(1/3,β). For β<1/3 that is σ+β+1/8, strictly smaller than the claimed σ+11/24. Example: β=0.2, σ=0.23 satisfies β+σ≥5/12 and σ<β+1/8, but the linear bound gives n^{-0.555}, not n^{-0.688}. The nonlinear sums have the same problem: (4.38b) and (4.38c) only give the exponents β+7/12 and β+1/8 when β≥1/4. So the induction closes only for β≥1/3; the theorem statement's β<1/3 range is unsupported.\n\nWhat is genuinely good: the spectral analysis is self-contained for convex/concave fluxes; Assumption 1 is stated honestly and Section 3.4 shows it can fail, which is the right kind of conditional result. The linear decay estimates are sharp and rest on prior published Green's function bounds [9], not on fitted assumptions. The numerical experiments are illustrative.\n\nWhere it is soft beyond the exponent issue: the general-flux results are conditional on Assumption 1, which is not flagged in the abstract; several proofs defer to 'similar arguments' or to [7] (Lemmas 5.1–5.3); Appendix A is heavy and hard to check in review. But the exponent mismatch in Chapter 5 is the load-bearing problem.\n\nWho should read this: anyone working on discrete shock profiles or dispersive numerical schemes. The linear and spectral parts will be useful even if the nonlinear theorem needs repair. It deserves a serious referee and a major revision: fix the parameter range (either restrict Theorem 2.5 to β≥1/3 or sharpen the semigroup estimate), then resubmit.\n\nRecommendation: send it to peer review. The result is significant enough and the linear analysis strong enough to warrant referee time, provided the referee checks the Chapter 5 exponents carefully.","headline":"Solid linear theory, overreaching nonlinear theorem: Theorem 2.5's ℓ∞ decay needs β≥1/3, not the stated β+σ≥5/12 range.","tokens_in":87306,"tokens_out":6807,"would_cite":true,"duration_ms":54023,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65M06","65M12","47B35","35L65","35L67"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that spectrally stable stationary shock profiles of the Lax-Wendroff scheme are nonlinearly orbitally stable: small perturbations converge to the shock profile carrying the perturbation's mass.","keywords":["hyperbolic conservation laws","shock waves","difference approximations","stability","Lax-Wendroff scheme","spectral stability","orbital stability","Green's function"],"falsifier":"Take the non-convex construction of Section 3.4 — a shock with $u_\\ell=1$, $u_r=-1$, flux values $f(\\pm1)=1$, and parameters $(\\alpha_\\ell,\\alpha_r)=(1/3,-2/3)$ with a small oscillation tuning $\\alpha_m\\approx 8.79$ — for which the paper's formulas give $\\Delta(2)=0$: running the scheme with a small perturbation should show growth rather than convergence, confirming that Assumption 1 is genuine. Alternatively, for any flux where sampling $\\Delta$ on the spectral curve shows no extra zeros, the theory predicts the exact algebraic decay rates $n^{-\\sigma}$ and $n^{-\\sigma-11/24}$; a numerical run deviating from these exponents would break the central claim.","tokens_in":85962,"feed_emoji":"🌊","tokens_out":13815,"duration_ms":112221,"temperature":0.7,"pith_summary":"The paper asks whether the Lax-Wendroff scheme, a second-order finite-difference method for hyperbolic conservation laws that produces spurious oscillations near shocks, can still keep stationary shock profiles stable. Its central claim is that spectral stability — the linearized scheme having no spectrum outside the unit disk except the simple eigenvalue $1$ — is sufficient for nonlinear orbital stability: every sufficiently small perturbation of the piecewise constant shock converges, at an algebraic rate, to a stationary shock profile whose mass equals the perturbation's mass. The result matters because the scheme's dispersive nature destroys the uniform $\\ell^1$ estimates on which classical parabolic stability proofs rely, so the stability question genuinely seemed open. The paper closes it by a sharp pointwise study of the Green's function, organized around an activation function that describes how the perturbation's mass accumulates along the eigenvector of the neutral eigenvalue.","feed_headline":"Lax-Wendroff shock profiles are orbitally stable","feed_subtitle":"A Green's-function analysis shows small perturbations decay to the same-mass shock despite dispersive oscillations.","key_machinery":"The load-bearing object is the Lopatinskii determinant $\\Delta(z)$, an explicitly computable holomorphic function that acts as a characteristic polynomial for the linearized scheme; it is assembled from the roots of the dispersion relations on the two sides of the shock. A simple zero at $z=1$ locates the neutral eigenvalue coming from the one-parameter family of profiles, and non-vanishing of $\\Delta$ on the closed exterior of the spectral ellipse of (2.13) is exactly spectral stability. From $\\Delta$ the paper builds the spatial Green's function, isolates its pole at $z=1$ with residue $H_j$ (the eigenvector), applies an inverse Laplace transform, and obtains the temporal Green's function whose leading term is the activation function $A(x,y)$ — a primitive of the free Green's function of the Lax-Wendroff transport equation — controlling how the mass of the initial condition accumulates along the eigenvector. The remainder terms are bounded by the functions $M_\\ell,M_r,K_\\ell,K_r$ encoding the oscillatory-diffusive profile of the dispersive scheme. Polynomial weights $\\ell^p_\\gamma$ enter because the free Green's function has the known sharp $n^{1/8}$ growth in $\\ell^1$, so the weights compensate the weak instability and make the nonlinear bootstrap close.","core_discovery":"Under the Rankine-Hugoniot relation, the Lax entropy inequalities $f'(u_r)<0<f'(u_\\ell)$, the CFL restriction $\\max(\\alpha_\\ell,|\\alpha_r|)<1$, and Assumption 1 on the zeros of the Lopatinskii determinant, the paper proves that the family of stationary discrete shock profiles $\\{v^\\theta\\}$ is nonlinearly orbitally stable. Concretely (Theorem 2.5), for any perturbation $h$ of the reference shock $u$ with $\\|h\\|_{\\ell^1_\\gamma}$ small enough, the numerical solution $u^n$ is well defined and satisfies $\\|p_n\\|_{\\ell^1_\\beta}\\le C_0(1+n)^{-\\sigma}\\|h\\|_{\\ell^1_\\gamma}$ and $\\|p_n\\|_{\\ell^\\infty_\\beta}\\le C_0(1+n)^{-\\sigma-11/24}\\|h\\|_{\\ell^1_\\gamma}$ for $p_n=u^n-v^\\theta$, with $\\theta=\\sum_j h_j$ and $\\gamma=\\sigma+\\beta+1/8$. The paper also proves that every convex or concave flux satisfies Assumption 1, so for such fluxes — Burgers's equation included — every Lax entropy shock is orbitally stable; and it exhibits non-convex fluxes for which Assumption 1 fails, showing the assumption is not vacuous.","pith_inferences":["The authors leave a direct test implicit: because $\\Delta$ is explicitly computable, Assumption 1 can be checked numerically for any proposed flux by sampling the determinant on the spectral curve; the non-convex examples of Section 3.4 show where such a check would find failures.","The activation-function machinery is likely the template for stability analysis of other higher-order dispersive schemes: the embedded neutral eigenvalue, the mass-accumulation function, and polynomial weights compensating dispersive growth should transfer, with exponents changing but the mechanism persisting.","One may read the sharp $n^{1/8}$ growth as intrinsic to second-order accuracy, suggesting a general principle that no uniform-in-time $\\ell^1$ stability theory can exist for such schemes, so weighted spaces are the natural habitat of their stability statements.","Appendix A's sharp bounds on the activation function supply the key estimate for a local limit theorem for dispersive finite-difference approximations of transport, a connection the paper flags as work in progress rather than develops."],"forward_implications":["For any convex or concave flux satisfying the Lax entropy inequalities and the CFL condition, stationary shock profiles of the Lax-Wendroff scheme are nonlinearly orbitally stable; the Burgers-equation shocks used in the numerical experiments are covered.","Conservation forces the selection of the limit: the asymptotic profile is the one whose excess mass equals the total mass of the initial perturbation, so stability holds for the whole one-parameter family $\\{v^\\theta\\}$, not for a single profile.","Decay is purely algebraic, at rates $n^{-\\sigma}$ in $\\ell^1_\\beta$ and $n^{-\\sigma-11/24}$ in $\\ell^\\infty_\\beta$, because the eigenvalue $1$ is embedded in the continuous spectrum and there is no spectral gap; exponential convergence cannot be expected in this setting.","The zero-weight case of the linear theory recovers the sharp $n^{1/8}$ growth of the Lax-Wendroff scheme in $\\ell^1$, confirming in the shock context that uniform $\\ell^1$ stability is impossible and that weighted spaces are the natural framework.","The constraint $\\beta+\\sigma\\ge 5/12$ with $0\\le\\sigma<\\beta+1/8$ quantifies exactly which polynomial weights allow the bootstrap argument to close."],"supporting_citations":[{"why":"Supplies the existence of the one-parameter family of stationary discrete shock profiles (Theorem 2.1) and the earlier stability framework that this paper extends.","marker":"[28]"},{"why":"Provides the sharp quantitative estimates on the Lax-Wendroff Green's function for the transport equation, on which the activation-function bounds and the linear analysis rest.","marker":"[9]"},{"why":"Identified and refined the slow ell-one growth of the Lax-Wendroff scheme that the paper's zero-weight estimates reproduce and that polynomial weights compensate.","marker":"[16, 17]"},{"why":"Supplies the general framework of discrete shock profiles, the Evans-function construction, and the definition of spectral stability used to formulate Assumption 1.","marker":"[27]"},{"why":"Establishes spectral stability in the symmetric case where the fluxes at the two end states have opposite derivatives, the starting point and a particular case of the present analysis.","marker":"[15]"},{"why":"Provides the nonlinear bootstrap strategy and the auxiliary lemmas on the nonlinear remainder that Chapter 5 adapts to the fully discrete setting.","marker":"[7]"}],"fun_headline_variants":["Shock profiles stay stable in Lax-Wendroff","Orbital stability proven for Lax-Wendroff shocks","Lax-Wendroff shock stability via dispersive Green's function","Convex fluxes guarantee stable Lax-Wendroff shocks","Lax-Wendroff: shock profiles resist small perturbations"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result rests on the premise, not proved in full generality, that the linearized scheme has no eigenvalue outside the unit disk except the neutral eigenvalue $1$ (a simple zero of the Lopatinskii determinant) — a condition the paper verifies only for convex or concave fluxes and shows can fail for non-convex ones.","fun_headline_variants_meta":{"raw":{"variants":["Shock profiles stay stable in Lax-Wendroff","Orbital stability proven for Lax-Wendroff shocks","Lax-Wendroff shock stability via dispersive Green's function","Convex fluxes guarantee stable Lax-Wendroff shocks","Lax-Wendroff: shock profiles resist small perturbations"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00026,"raw_usage":{"total_tokens":1686,"prompt_tokens":1141,"completion_tokens":545,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":757,"completion_tokens_details":{"reasoning_tokens":461}},"tokens_in":757,"tokens_out":545,"duration_ms":30137,"temperature":1.0,"reasoning_tokens":461,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:51:55.978935+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the non-convex construction of Section 3.4 — a shock with $u_\\ell=1$, $u_r=-1$, flux values $f(\\pm1)=1$, and parameters $(\\alpha_\\ell,\\alpha_r)=(1/3,-2/3)$ with a small oscillation tuning $\\alpha_m\\approx 8.79$ — for which the paper's formulas give $\\Delta(2)=0$: running the scheme with a small perturbation should show growth rather than convergence, confirming that Assumption 1 is genuine. Alternatively, for any flux where sampling $\\Delta$ on the spectral curve shows no extra zeros, the theory predicts the exact algebraic decay rates $n^{-\\sigma}$ and $n^{-\\sigma-11/24}$; a numerical run deviating from these exponents would break the central claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the existence of the one-parameter family of stationary discrete shock profiles (Theorem 2.1) and the earlier stability framework that this paper extends."},{"cited_title":"Coulombel","cited_arxiv_id":null,"evidence_quote":"Provides the sharp quantitative estimates on the Lax-Wendroff Green's function for the transport equation, on which the activation-function bounds and the linear analysis rest."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the general framework of discrete shock profiles, the Evans-function construction, and the definition of spectral stability used to formulate Assumption 1."},{"cited_title":"Harten, J","cited_arxiv_id":null,"evidence_quote":"Establishes spectral stability in the symmetric case where the fluxes at the two end states have opposite derivatives, the starting point and a particular case of the present analysis."},{"cited_title":"Nonlinear orbital stability of stationary discrete shock profiles for scalar conservation laws","cited_arxiv_id":"2409.18930","evidence_quote":"Provides the nonlinear bootstrap strategy and the auxiliary lemmas on the nonlinear remainder that Chapter 5 adapts to the fully discrete setting."}],"review_version":1}