{"id":"8fc8908e-32d2-44fb-8319-b614d8016a7c","arxiv_id":"1908.01632","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For fractional Laplacian order 1 < alpha < 2, solutions of the scaled fractal Burgers equation converge in L^2 to the entropy shock, with a shift and a convergence rate that tends to zero.","lead":"This math paper proves a quantified rate for the vanishing viscosity limit of the fractal Burgers equation toward entropy shocks, allowing large initial perturbations around the shock. It extends the relative entropy method from the classical Laplacian to fractional dissipation in one space dimension.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 2.6 uses the invalid inequality √(δεβ+E) ≤ √δεβ+E, so the stated rate ψ is not established.","rationale":"The reader's CONDITIONAL verdict is appropriate. The most load-bearing issue is not the unquantified tail of S1 (which weakens the sharpness of the rate but not its existence) but the invalid square-root linearization in §2.6, which directly invalidates the stated quantitative rate. This is a concrete algebraic error rather than a gap in the relative-entropy framework: Propositions 2.1–2.3 appear internally coherent under the paper's regularity assumptions. The proof can likely be repaired by defining ψ with √E in place of E, which still vanishes as ε→0. Because the central convergence claim survives qualitatively but the theorem's explicit formula is not established, no change from the reader's conditional verdict is needed.","tokens_in":13491,"tokens_out":25165,"duration_ms":278875,"concrete_test":"Re-derive the chain in §2.6 keeping the exact square root: the final constant must contain √E, not E. Then test the stated ψ numerically for ε=10^{−6}, α=3/2 (so β=2), and δ=ε^{−β/2}: compute √(δεβ+E) and compare with √δεβ+E. If the former exceeds the latter, the theorem's ψ cannot be an upper bound. A corrected rate with √E still tends to 0, so the concern is fixable without changing the qualitative result.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof's last estimate in Section 2.6 is invalid as written. After bounding the unweighted L2 difference by ||u0−Sε|| + C(T)√(δεβ+E), the text continues with ||u0−Sε|| + C(T)(√δεβ + E + εβ/2). This uses √(a+E) ≤ √a+E, which fails whenever 0 < E < 1 (e.g., a=0, E=0.01). The quantity E defined in §2.6 is √δεβ + (S1(√δ)−u+) + (u− − S1(−√δ)) + (1/δ)^{α−1}, and Remark 1.2 shows E→0 as ε→0 by choosing δ=ε^{−β/2}. Thus the regime where the inequality fails is exactly the regime needed for convergence. Consequently, the explicit formula for ψ(ε) in Theorem 1.1 is not derived from the preceding bounds. The main idea is not destroyed: the valid inequality √(δεβ+E) ≤ √δεβ + √E would give a different, still vanishing rate containing square roots of the layer-tail terms. But the theorem as stated, with ψ containing E linearly, is not proven.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the vanishing fractal-viscosity limit for the one-dimensional fractal Burgers equation ∂_t u + ∂_x(u^2/2) = ε Δ^{α/2}u with 1 < α < 2. The main result, Theorem 1.1, asserts an L^2 convergence estimate, up to a Lipschitz shift X_ε(t), from solutions of the scaled equation to the inviscid entropy shock S_0, for initial data that are L^∞ with u_0 - S_0 ∈ L^2 and (u_0')_+ ∈ L^2. The rate is encoded in a function ψ(ε) built from the shock layer S_1 of the stationary profile. The proof combines the relative entropy method with a shift ODE taken from Leger's normalized relative flux, a monotonicity lemma for the layer, a one-sided derivative bound, and a nonlocal parabolic estimate.","tokens_in":13658,"tokens_out":15770,"duration_ms":131726,"significance":"If the stated result holds, it would be a valuable extension of the relative-entropy inviscid-limit theory to nonlocal viscosity for large perturbations, and several components of the proof are correct and useful: the monotonicity of the shock layer in Lemma 2.1, the positive-part derivative bound in Lemma 2.2, the shift construction, and the nonlocal parabolic estimate in Proposition 2.3. However, the final rate as stated is not established because of an invalid inequality in the concluding step, so the paper requires a nontrivial correction before the central quantitative claim can be accepted.","major_comments":[{"comment":"The proof obtains the valid bound ‖u_ε(·+X(t),t)-S_ε‖_{L^2} ≤ ‖u_0-S_ε‖_{L^2} + C(T)√(δε^β+E(ε,δ)), where E(ε,δ)=√(δε^β)+(S_1(√δ)-u_+)+(u_- - S_1(-√δ))+(1/δ)^{α-1}. The next displayed step replaces C(T)√(δε^β+E) by C(T)(√(δε^β)+E). This uses the inequality √(a+b) ≤ √a + b, which is false whenever 0 < b < 1. Since Remark 1.2 chooses δ=ε^{-β/2} precisely to make E(ε,δ)→0, the inequality fails in exactly the regime needed for convergence. The formula for ψ(ε) displayed in Theorem 1.1, with two linear √(δε^β) terms and E appearing linearly, is precisely the outcome of this invalid step, so the stated rate is not derived. The error is repairable: using the valid bound √(δε^β+E) ≤ √(δε^β)+√E would give a still-vanishing rate, with square roots of the layer-tail terms, and would preserve the qualitative conclusion. But as written, the central quantitative claim of the paper is not proved.","section":"§2.6, final display; Theorem 1.1"}],"minor_comments":[{"comment":"The definition of ψ(ε) refers to E(ε,δ) even though E is not defined until §2.6, and the displayed expression with two √(δε^β) terms is inconsistent with the proof's final formula. The theorem statement should define all quantities explicitly and match the bound actually proved.","section":"Theorem 1.1"},{"comment":"The word 'quantified' in the abstract and Theorem 1.1 is stronger than what is delivered, since ψ(ε) depends on the layer-tail values S_1(√δ)-u_± whose decay rate is unknown and is explicitly said to be unknown in Remark 1.1. The authors should rephrase the claim, for example as 'a rate expressible in terms of the layer tail', unless an independent estimate for that tail is supplied.","section":"Remark 1.1; Theorem 1.1"},{"comment":"The notation for the localization weight is inconsistent across sections: Proposition 2.1 writes φ_δ(|x|/ε^β), while after the change of variables in Proposition 2.2 the weight is written φ_δ(|x|). Defining the rescaled weight once and then referring to it would improve readability.","section":"§2.2 and §2.4"},{"comment":"In the estimate of h_1, the notation v_ε(0,t) = u_ε(X(t),t) is used without restating the definition from (2.13); a short reminder would help the reader follow the shift-dependent argument.","section":"§2.3, proof of Proposition 2.1"}],"recommendation":"major_revision","confidential_remarks":"The flaw is localized to the final step of the proof and is fixable, while the main estimates appear sound. I would encourage the authors to repair the final inequality, restate ψ accordingly, and then resubmit. The current theorem statement should not appear as is."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Good to see this paper out. You should know two things. The main advance is real: it genuinely extends Choi–Vasseur to the nonlocal fractional Laplacian for large L2 perturbations, and the nonlocal parabolic estimate (Prop. 2.3) is solid new work. But the last estimate in Section 2.6 has a genuine bug: they move from sqrt(δεβ + E) to sqrt(δεβ) + E, which fails when E is small and positive. Since E is exactly the small parameter that tends to zero, the stated ψ(ε) is not actually derived. That doesn't kill the argument: the valid inequality sqrt(a+E) ≤ sqrt(a)+sqrt(E) gives a different still-vanishing rate with square roots of the layer-tail terms. So it's a patch, not a hole in the strategy.\n\nThe rest of the proof holds up. Lemma 2.1 (monotonicity of the layer) and Lemma 2.2 (one-sided derivative bound) are clean. The cutoff argument with φδ is careful. The authors are honest in Remark 1.1 that the layer decay rate is unknown, which means the 'quantified' rate is not a clean power law; the abstract oversells it a bit. Self-citations to Choi–Vasseur and Kang–Vasseur are legitimate; they are prior results, not a fit.\n\nMy take: this deserves a serious referee and likely publication after a straightforward correction. The flawed step is easy to fix, and the main theorem still holds in a slightly modified form. I'd send it to review rather than desk reject, with the instruction to fix the sqrt inequality and make the rate claim match what is actually proven.\n\nFor your reading group, it's worth a session: the nonlocal estimates are instructive, and the final slip is a nice lesson in why inequality steps need checking.","headline":"Genuine extension of the relative entropy method to the fractional Laplacian, but a real inequality slip in the final step undermines the stated rate.","tokens_in":14254,"tokens_out":2587,"would_cite":true,"duration_ms":25282,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35L65","35L67","35B35","35B40"],"pacs":[],"model":"deepseek-v4-flash","headline":"Entropy shocks of the fractal Burgers equation are stable in the vanishing-viscosity limit with a quantified rate, even for large perturbations.","keywords":["fractal Burgers equation","fractional Laplacian","inviscid limit","entropy shock","relative entropy method","shock layer","large perturbation","vanishing viscosity"],"falsifier":"Take $\\alpha=3/2$, $u_-=1$, $u_+=-1$, and initial data $u_0=S_0+\\chi_{[-1,1]}$ after rescaling so that $\\|u_0-S_0\\|_{L^2}=1$; compute numerically, at $t=1$ and $\\varepsilon=10^{-2},10^{-3},\\ldots$, the $L^2$ error between $u_\\varepsilon(\\cdot+X(t),t)$ and $S_0$ using the shift defined by the paper's ODE. If the quantity $\\|u_\\varepsilon(\\cdot+X(t),1)-S_0\\|_{L^2}-\\|u_0-S_0\\|_{L^2}$ does not tend to zero as $\\varepsilon\\to0$, Theorem 1.1 is false.","tokens_in":13219,"feed_emoji":"🌊","tokens_out":18475,"duration_ms":152622,"temperature":0.7,"pith_summary":"This paper proves the first quantitative vanishing-viscosity limit for entropy shocks of the fractal Burgers equation $u_t + (u^2/2)_x = \\varepsilon \\Delta^{\\alpha/2}u$ in the range $1<\\alpha<2$. It shows that for any initial datum that is an $L^2$ perturbation of an entropy shock and whose derivative has a positive part in $L^2$, the viscous solution, after shifting the spatial coordinate by a Lipschitz function $X_\\varepsilon(t)$, stays within distance $\\|u_0-S_0\\|_{L^2} + C(T)\\psi(\\varepsilon)$ of the inviscid shock on any bounded time interval, with $\\psi(\\varepsilon)\\to0$ as $\\varepsilon\\to0$. The significance is that the perturbation may be large, so the shock layer is destroyed in the limit, yet the entropy shock is still recovered. The rate $\\psi(\\varepsilon)$ is explicit in terms of the dissipation parameters and the tail of the viscous shock layer, though the layer tail's own decay rate is left open.","feed_headline":"Fractal Burgers shocks now come with a quantified error","feed_subtitle":"A shift absorbs the phase error; even large perturbations add only a vanishing excess over the initial error.","key_machinery":"The mechanism is the relative entropy method with a localized weight. The paper tracks $H(t)=\\int_{\\mathbb{R}}\\phi_\\delta^2(|x|/\\varepsilon^\\beta)\\frac{|u_\\varepsilon(x+X(t),t)-S_\\varepsilon(x)|^2}{2}\\,dx$, where $S_\\varepsilon(x)=S_1(x/\\varepsilon^\\beta)$ is the monotone shock layer of the scaled equation, $\\phi_\\delta$ is a smooth cutoff that localizes to a window of width $\\delta\\varepsilon^\\beta$, and the shift $X(t)$ is chosen to solve $\\dot X=f(u_\\varepsilon(X(t),t),S_\\varepsilon(0))$, with $f$ the normalized relative entropy flux from [20]. The choice of shift cancels the main hyperbolic term; the remaining three contributions are bounded by $C\\sqrt{\\delta\\varepsilon^\\beta}$ (hyperbolic term from the cutoff), by $C((1/\\delta)^{3/2}+(S_1(\\sqrt\\delta)-u_+)+(u_- - S_1(-\\sqrt\\delta)))$ (a second hyperbolic term controlled by the monotonicity of the layer), and by $C(1/\\delta)^{\\alpha-1}$ (the fractional dissipation term). Minimizing over $\\delta\\ge4$ yields $\\psi(\\varepsilon)$, and the layer's monotonicity, proved from the strict convexity of the flux and the structure of the fractional Laplacian, is what makes the middle estimate possible.","core_discovery":"On the paper's own terms, the central result is Theorem 1.1: for $1<\\alpha<2$, any end states $u_->u_+$, and initial data $u_0$ with $u_0-S_0\\in L^2(\\mathbb{R})$ and $(\\partial_x u_0)_+\\in L^2(\\mathbb{R})$, for every $T>0$ there exists a Lipschitz shift $X_\\varepsilon$ with $X_\\varepsilon(0)=0$ such that for all $t\\le T$, \\[\\|u_\\varepsilon(\\cdot+X_\\varepsilon(t),t)-S_0(\\cdot-\\$\\sigma$ t)\\|_{$L^{2}$(\\mathbb{R})}\\le \\|u_0-S_0\\|_{$L^{2}$(\\mathbb{R})}+C(T)\\psi(\\varepsilon),\\] where $\\psi(\\varepsilon)\\to0$. This is the first result of this kind for the fractal Burgers equation: previous convergence statements either used no initial perturbation or restricted to small perturbations, whereas here the perturbation can be arbitrarily large in $L^2$. A shift is necessary, since $L^2$ contraction to the shock without a shift is already known to fail. The proof establishes the same estimate for convergence to the smooth monotone shock layer $S_\\varepsilon(x)=S_1(x/\\varepsilon^\\beta)$ of width $\\varepsilon^\\beta$, and then uses the $L^2$ convergence of the layer to the discontinuous shock.","pith_inferences":["The shift $X_\\varepsilon$ is not a nuisance: it carries the phase drift caused by the excess entropy of a large perturbation, and the same shift-based cancellation may be the right framework for other nonlocal conservation laws with monotone traveling waves.","A sharper closed-form estimate for the tail of $S_1$ at specific $\\alpha$ would immediately upgrade $\\psi(\\varepsilon)$ to an explicit power of $\\varepsilon$; this is the natural next computation.","The hypothesis $(\\partial_x u_0)_+\\in L^2$ is a one-sided regularity condition that could perhaps be relaxed, but testing large oscillatory data with unbounded positive slope would show whether the method's reliance on this bound is essential.","As $\\alpha\\to2^-$, $\\beta=1/(\\alpha-1)\\to\\infty$, so the layer width is extremely small; comparing the rate with the local Laplacian case suggests the large-perturbation statement is most delicate near $\\alpha=2$."],"forward_implications":["For every $1<\\alpha<2$ and every bounded time interval, the entropy shock is stable in $L^2$ under the vanishing-viscosity evolution for large perturbations, with the spatial shift absorbing the phase error that would otherwise make contraction fail.","When the shock layer decays exponentially, the excess error is at most $C\\varepsilon^{1/(2(2\\alpha-1))}$, a rate only slightly worse than the optimal layer rate $\\varepsilon^{1/(2(\\alpha-1))}$.","The argument is written for a general strictly convex flux $A$, so any scalar conservation law with a convex nonlinearity and a monotone fractional shock layer inherits the same quantitative stability statement.","If the initial datum is exactly the scaled layer, the rate matches the optimal layer rate, showing the theorem is sharp up to the unresolved layer-tail contribution."],"supporting_citations":[{"why":"[10] supplies global smooth solutions and the $L^\\infty$ maximum principle for the scaled fractal Burgers equation, which are used throughout the estimates.","marker":"[10]"},{"why":"[5] proves the existence of the smooth shock layer $S_1$ for $1<\\alpha<2$; the layer approximation and the rate $\\psi(\\varepsilon)$ depend on it.","marker":"[5]"},{"why":"[6] is the local-Laplacian predecessor whose hyperbolic estimates and derivative lemma are adapted here to the nonlocal case.","marker":"[6]"},{"why":"[20] defines the normalized relative entropy flux and its monotonicity properties used to construct the shift, and it shows that contraction without a shift is false.","marker":"[20]"},{"why":"[16] supplies the relative-entropy identity for viscous shocks and the exponential-layer comparison used in the optimal-rate discussion.","marker":"[16]"}],"fun_headline_variants":["Fractal Burgers: quantified shock limit for large perturbations","First quantified inviscid limit for fractal Burgers entropy shocks","Large L2 perturbations: quantified shock convergence for fractal Burgers","Shifted error bound for fractal Burgers shocks with any L2 data","Inviscid limit rate for fractal Burgers: large perturbations allowed"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof needs a smooth monotone shock layer $S_1$ for the stationary fractional Burgers equation, and the claimed rate inherits the unknown rate at which that layer approaches the end states $u_\\pm$; if the layer tail decays very slowly or the layer is not monotone, the estimate loses its rate or its control of the hyperbolic term.","fun_headline_variants_meta":{"raw":{"variants":["Fractal Burgers: quantified shock limit for large perturbations","First quantified inviscid limit for fractal Burgers entropy shocks","Large L2 perturbations: quantified shock convergence for fractal Burgers","Shifted error bound for fractal Burgers shocks with any L2 data","Inviscid limit rate for fractal Burgers: large perturbations allowed"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000599,"raw_usage":{"total_tokens":2779,"prompt_tokens":902,"completion_tokens":1877,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":518,"completion_tokens_details":{"reasoning_tokens":1788}},"tokens_in":518,"tokens_out":1877,"duration_ms":11323,"temperature":1.0,"reasoning_tokens":1788,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:08:36.302707+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $\\alpha=3/2$, $u_-=1$, $u_+=-1$, and initial data $u_0=S_0+\\chi_{[-1,1]}$ after rescaling so that $\\|u_0-S_0\\|_{L^2}=1$; compute numerically, at $t=1$ and $\\varepsilon=10^{-2},10^{-3},\\ldots$, the $L^2$ error between $u_\\varepsilon(\\cdot+X(t),t)$ and $S_0$ using the shift defined by the paper's ODE. If the quantity $\\|u_\\varepsilon(\\cdot+X(t),1)-S_0\\|_{L^2}-\\|u_0-S_0\\|_{L^2}$ does not tend to zero as $\\varepsilon\\to0$, Theorem 1.1 is false.","supporting_citations":[{"cited_title":"Droniou, T","cited_arxiv_id":null,"evidence_quote":"[10] supplies global smooth solutions and the $L^\\infty$ maximum principle for the scaled fractal Burgers equation, which are used throughout the estimates."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"[5] proves the existence of the smooth shock layer $S_1$ for $1<\\alpha<2$; the layer approximation and the rate $\\psi(\\varepsilon)$ depend on it."},{"cited_title":"Choi and A","cited_arxiv_id":null,"evidence_quote":"[6] is the local-Laplacian predecessor whose hyperbolic estimates and derivative lemma are adapted here to the nonlocal case."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"[20] defines the normalized relative entropy flux and its monotonicity properties used to construct the shift, and it shows that contraction without a shift is false."},{"cited_title":"Kang and A","cited_arxiv_id":null,"evidence_quote":"[16] supplies the relative-entropy identity for viscous shocks and the exponential-layer comparison used in the optimal-rate discussion."}],"review_version":1}