{"id":"92ce3aa1-cb61-4951-bbcf-7674ca570920","arxiv_id":"2412.12530","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A rigorous Bäcklund transform is constructed for KP-II on the plane, and its image is shown to be a codimension-1 manifold, yielding codimension-1 L2 stability of the line soliton at sharp regularity.","lead":"The paper builds a Bäcklund transform that adds a line soliton to solutions of the KP-II water-wave equation, and shows a scalar constraint must hold for the addition to work. This yields a codimension-1 L2 stability result for the line soliton and a new route toward multisoliton stability.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The uniqueness claim in Theorem D is not established: its proof assumes half-plane L3,∞ tails vanish, a property that is false for this function class.","rationale":"The reader identified the same load-bearing assumption: the uniqueness claim in the proof of Theorem D for solutions of the forced Burgers equation (5.2) in L3,∞ ∩ cosh(x)L2. My stress-test sharpens this into a concrete failure at the level of the proof: the half-plane smallness of w1+w2 in X is asserted from membership in X, but membership in L3,∞ ∩ L2 (or the weighted L2 space) does not imply vanishing L3,∞ tails at y=-∞. The separated-block counterexample shows the general statement is false, so the proof as written has a genuine gap, not merely a compressed estimate. This gap is load-bearing because the contradiction in the 'only if' direction of Theorem D depends on it. I did not find a separate flaw in the forward direction Φ(g)=0 ⇒ in range, nor in the L2 stability argument modulo that forward direction; those parts may be salvageable. Therefore I do not move the reader's verdict away from CONDITIONAL: the manuscript should be accepted only after this uniqueness/tail-decay step is either repaired with an argument that uses equation (5.2) itself, or replaced by a counterexample that would invalidate Theorem D.","tokens_in":74935,"tokens_out":13197,"duration_ms":128228,"concrete_test":"Verify the tail-smallness assertion with f_M := Σ_{n>M} n 1_{[-n^{-4},0]×[-n,-n+1]}. If ‖f_M‖_{L3,∞} does not tend to 0 as M→∞ (it is ≳1), the proof's half-plane smallness step is invalid for general X. Then test the uniqueness claim directly: for g=0 in (5.2), seek nonzero solutions of the form w(y,x)=e^{λy}φ(x) in L3,∞ ∩ cosh(x)L2 by looking for L2 eigenfunctions of the x-operator -∂xx - 2∂x(tanh·). Finding a nonzero φ would disprove the Claim and hence the 'only if' of Theorem D; proving nonexistence of such φ would indicate the theorem may be recoverable by a different argument.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The 'only if' direction of Theorem D rests on the Claim in §5.6 that two solutions w1,w2 of (5.2) in L3,∞(R2) ∩ cosh(x)L2(R2) with the same forcing must coincide. The proof's only mechanism for this is the assertion that, for M negative enough, 1_{yα}| ≈ Σ_{n≥α} n^{-4} ≈ α^{-3}, so α |{|f|>α}|^{1/3} = O(1)). Membership in L^{3,∞} ∩ cosh(x)L2 therefore imposes no decay in y. The same gap invalidates the claim that a homogeneous linear solution 'must be zero by sending y0 to −∞', which presupposes ‖z(y0)‖_{L1+L∞} → 0 as y0 → −∞. Because this uniqueness is exactly what forces w = w′ when Φ(g) ≠ 0, the 'only if' direction of Theorem D is unproved as written. Corollary E is not affected: its stability statement uses only the forward implication Φ(g) = 0 ⇒ g + φ in the range of B, which is independent of the Claim.","section":"Section 5.6 (Claim in the proof of Theorem D)"},{"comment":"The strict non-integrability assertion 'ψx, w ∈ L^{3,∞}(R2) \\ L3(R2)' when Φ(g) ≠ 0 is supported only by 'a quick study of the kernel of Γ− − Γ+'. This assertion is load-bearing: it is the property that contradicts w′ ∈ L3(R2) at the end of the proof. The text establishes the weak-L³ membership (via Γ± ∈ L^{3,∞} and sech²hψ ∈ L1) but not the failure of the L3-integrability, which requires a quantitative lower bound on the singular part of II = ½ ∂x^{-1}(Γ− − Γ+)(sech²hψ). As printed this is a plausibility argument, not a proof; and even after repair, the contradiction is complete only once the uniqueness Claim of Major Comment 1 is available.","section":"Section 5.6 (proof of Theorem D, second half)"},{"comment":"The weighted-estimate step in the Hardy-space check contains an arithmetic inconsistency. The text sets 1/p = (1−θ)/1 + θ/2 and states that '1 > θ > 3/(3+2ε) ⇔ δ := θε − 3/p′ > 0'. From the displayed interpolation relation one obtains p = 2/(2−θ), hence p′ = 2/θ and 3/p′ = 3θ/2, so δ = θ(ε − 3/2). Thus δ > 0 requires ε > 3/2, and the displayed threshold θ > 3/(3+2ε) is not equivalent (it corresponds to θε > 3(1−θ)/2). Since Theorem D and Corollary E are stated for every ε > 0, the proof as written does not cover ε ≤ 3/2. This issue is localized and likely repairable — for example by a direct Lp estimate using the (1+|y|)^ε decay of the data, or by restricting the statements to ε > 3/2 — but as printed it is an error.","section":"Section 5.6 (Theorem D, forward implication)"}],"minor_comments":[{"comment":"The clause 'which the potential for generalization to multisolitons' is grammatically incomplete in both the abstract and the introduction; it should presumably read 'with the potential for generalization to multisolitons'.","section":"Abstract and Section 1"},{"comment":"'Polinomially localized perturbations' should be 'polynomially localized perturbations'.","section":"Section 5.7"},{"comment":"'Derivating in x' should be 'differentiating with respect to x'; the same typo appears twice in the second half of the proof of Theorem D.","section":"Section 5.6"},{"comment":"The entries 'The map V is a generalization of V up to a change of variables' and 'The map V→→ is a time-dependent version of V' use indistinguishable symbols; the intended notation with the vector parameter ⃗λ (as in Propositions 3.19 and 4.17) should be restored.","section":"Cross-reference list in Section 1.4"},{"comment":"The large-data uniqueness argument is compressed ('The uniqueness for large v ∈ L3 is a consequence of the uniqueness for small v, translation invariance, and the fact that ...'); since this uniqueness underpins the elementary solutions v± used throughout Sections 3 and 4, a fuller statement or a reference would improve verifiability.","section":"Proof of Lemma 3.1"},{"comment":"Estimate (5.6) depends on the bound ‖Γ±(x,·)‖_{L2(Ry)} ≲ log^{1/2}(|x|) + ⟨x⟩^{-1/4}, which is stated without derivation; a short proof would help the reader verify the weighted claim.","section":"Section 5.5"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is ambitious, long, and in many parts well executed, and the main theorems are stated with unusual honesty about their limitations. My principal concern is the Claim in Section 5.6: the uniqueness of solutions of (5.2) in L^{3,∞} ∩ cosh(x)L2. The proof given is not valid, and I would recommend that, before acceptance, a careful referee verify either a correct proof of that claim (for instance via additional a priori decay in y that follows from the equation) or a reformulation of Theorem D in a class that encodes y-decay. If the claim turns out to be false, the range characterization would be one-sided ({Φ = 0} contained in the range) rather than exact; the paper's stability corollary (Corollary E) and commutation theorem (Theorem C) would survive. The ε-range inconsistency in Section 5.6 is minor by comparison but should be corrected. The paper's length and the number of typos also suggest a careful final pass."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take on Pompili's KP-II Bäcklund paper. The genuinely new thing is the construction of the Miura-map-based Bäcklund transform for KP-II on R^2, and the proof that it commutes with the flow modulo a time-dependent γ0. That's a real piece of work, and the classification of eternal solutions to the forced viscous Burgers equation via Cole-Hopf and parabolic BMO is careful and mostly convincing. The forward half of Theorem D — Φ(g)=0 implies the perturbation lies in the range of B — looks in good shape, and the resulting codim-1 L2 stability corollary at sharp regularity would be a solid contribution if the forward half is airtight.\n\nThe soft spot is the reverse half of Theorem D, specifically the uniqueness claim in §5.6. The proof needs that two solutions of (5.2) in L3,∞ ∩ cosh(x)L2 with the same forcing coincide. The mechanism is a bootstrap on half-planes y