{"id":"cbdb651c-9e19-4262-998d-f913762b11f0","arxiv_id":"1908.08768","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Periodic multi-soliton solutions of KdV of any size persist, with linear stability, under small smooth quasi-linear Hamiltonian perturbations.","lead":"The paper proves that large, quasi-periodic multi-soliton solutions of the KdV equation survive small quasi-linear Hamiltonian perturbations. It is the first KAM result for strongly nonlinear perturbations of an integrable PDE at arbitrary amplitude.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central proof hinges on Theorem 3.2 from companion [19]; if its pseudo-differential expansions or negative-Sobolev extensions fail, the KAM reduction in Sections 6–7 collapses.","rationale":"I read the strongest claim as the persistence of arbitrary-size finite-gap KdV tori under quasi-linear Hamiltonian perturbations, proved by Nash-Moser with a KAM reducibility step. The condition that must hold for the central claim is Theorem 3.2: near a compact family of S+-gap potentials there exists a canonical chart with pseudo-differential expansions of the map, its transpose, and their differentials, with tame estimates valid down to negative Sobolev orders. The manuscript explicitly defers this theorem to the companion paper [19], and the present text uses it as a black box in Sections 3, 5, 6, and 7. This is not an internal inconsistency, and I found no other step where the argument appears to break. The concern is therefore the same as the reader's weakest assumption: the main theorem is conditional on the correctness of [19]. Since the reader already made the verdict CONDITIONAL on this basis, my read does not change the verdict. The concrete test I propose is an independent verification of Theorem 3.2 in the simplest nontrivial case, which would settle whether the dependency actually lands.","tokens_in":87213,"tokens_out":6388,"duration_ms":74249,"concrete_test":"Verify Theorem 3.2 independently in the simplest non-trivial case S+={1}: starting from the global Birkhoff coordinates of [20], construct the chart Ψν of [19] near a one-gap torus and compute the expansion (3.5) and its transpose (3.6) to order M=2, checking the identity a_{-1}^{Ψ⊤}=-a_{-1}^{Ψ}, the claimed remainder order -M-1, and the tame estimates (Est1)–(Est2) for the negative-order extensions. If any of these checks fails, Sections 6–7 and hence Theorem 1.1 fail; if they pass, the conditional dependency is satisfied.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The entire KAM/Nash-Moser construction is built on the canonical chart Ψν whose existence and pseudo-differential expansion (3.5)–(3.6), up to remainders of arbitrary negative order with tame estimates, is imported as Theorem 3.2 from [19]. This manuscript proves consequences (Corollaries 3.3 and 3.4) and uses them in an essential way: those corollaries provide the negative-order extensions of dΨν and d⊥Ψν that feed into Lemmas 2.24, 3.5, and 3.7, and hence into the expansion of Lω in Lemma 6.3 and the subsequent reduction in Sections 6.3–6.5 and the KAM scheme of Section 7. If any of the claimed expansions—especially the transpose identity (3.7), the remainder order in (3.6), or the tame estimates (Est1)–(Est2)—were off by even one derivative, the order of the leading pseudo-differential part of Lω would change, the highest-order coefficient a3 would not be controllable, and the whole reducibility and inversion argument would fail. No internal contradiction is apparent, and the companion paper is published, so this is a structural dependency rather than a demonstrated flaw; but it is exactly the premise on which the central claim rests.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proves Theorem 1.1: for any finite subset S+ of positive integers and any Hamiltonian perturbation of KdV of the form ∂t u = ∂x ∇Hε(u) with Hε = Hkdv + εP and P(u) = ∫ f(x,u,ux)dx, f ∈ C∞, there is, for sufficiently small ε, a set of parameters of asymptotically full measure on which the perturbed equation admits a quasi-periodic solution close to the corresponding S+-gap solution of KdV, with frequency close to -ωkdv, and with a linearly stable invariant torus. The proof combines the canonical coordinates of KdV with the pseudo-differential normal-form chart Ψν constructed in the companion paper [19], derives pseudo-differential expansions for the linearized Hamiltonian vector fields, reduces the linearized operator Lω to a constant-coefficient operator of order three up to order-zero remainders, and then applies a KAM reducibility scheme and a Nash-Moser iteration with measure estimates.","tokens_in":87442,"tokens_out":9701,"duration_ms":101772,"significance":"If correct, the theorem resolves a longstanding question: it gives KAM persistence of finite-gap solutions of arbitrary size under genuinely quasi-linear Hamiltonian perturbations, going substantially beyond the earlier semilinear and small-amplitude results. The paper is technically very substantial: it contains detailed tame estimates, a quantitative Egorov theorem for transport flows, a careful reduction of Lω through four changes of variables, a modulo-tame KAM reducibility scheme, and explicit measure estimates based on the non-degeneracy of KdV frequencies. The dependence on the companion paper [19] is stated transparently, and the results of Sections 3.2-8 are largely self-contained once Theorem 3.2 is granted.","major_comments":[{"comment":"Theorem 3.2 is the load-bearing premise of the entire paper: the pseudo-differential expansions (3.5)-(3.6), the normal-form property (AE3), and the tame estimates (Est1)-(Est2) are imported from the companion paper [19]. Corollaries 3.3 and 3.4, Lemmas 3.5 and 3.7, and through them Lemma 6.3 and the whole reduction in Sections 6-7, all depend on this theorem. The manuscript proves consequences of Theorem 3.2 but not the theorem itself. Because [19] is published, this is a structural dependency rather than an internal inconsistency; nonetheless, the referee cannot verify the central claim from this manuscript alone. The authors should state explicitly which assertions of [19] are being used in exactly what form and confirm that no modification of the remainder orders or tame estimates of [19] is needed, since a change of even one derivative in (3.5)-(3.6) would alter the order of the leading pseudo-differential part of Lω and invalidate the reducibility argument.","section":"§3.1, Theorem 3.2"}],"minor_comments":[{"comment":"The text contains several typos and small grammatical errors, for example 'noe' after (5.3), 'inveritibility' near (5.22), 'Correpondingly' in Section 2.1, and 'repsectively' in the introduction; these should be corrected.","section":"§5, after (5.3)"},{"comment":"The operator Lω^(1) is called a Hamiltonian operator after conjugation by the time reparametrization Φ^(1) and scaling by 1/ρ. Since Φ^(1) is not a symplectic map of the phase space, the manuscript should explicitly justify why the conjugated and scaled operator remains Hamiltonian (for instance by noting that ρ is independent of x and that Φ^(1) commutes with ∂x), because the later KAM scheme uses the Hamiltonian structure to ensure real frequencies and the symmetry properties of the diagonal part.","section":"§6.2, Eq. (6.20)"},{"comment":"The proof of Lemma 2.6 uses formal manipulations with the nonlocal operator ∂x^{-1}; the statement would be clearer if it noted that the expansion is an asymptotic pseudo-differential expansion, so that the computation of the leading coefficients is justified in the symbol calculus rather than by exact operator identities.","section":"§2.2, Lemma 2.6"},{"comment":"In the display (6.9), the term Op(r0^(0)) is said to have an expansion (6.11) in homogeneous components, but the remainder after the sum in (6.11) is not shown explicitly; adding the remainder term there would make the statement consistent with the later use of the expansion in the Egorov reductions.","section":"§6.1, Lemma 6.3"}],"recommendation":"minor_revision","confidential_remarks":"The paper is a direct continuation of [19], which is by two of the same authors and is published. The referee should be aware that the correctness of Theorem 1.1 is conditional on the full strength of Theorem 3.2 of [19]; this is acceptable if that companion paper is available to the reader, but it does mean the present manuscript is not self-contained at its most critical point. No other substantive concerns emerged from my reading."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—this paper is the first large-KAM-tori result for quasi-linear perturbations of an integrable PDE. That is a genuine threshold. The theorem: for any smooth Hamiltonian perturbation with f(x,u,ux), finite-gap tori of arbitrary size persist for most parameters, with linear stability. The proof is a Nash-Moser scheme with pseudo-differential reduction of the linearized operator, Egorov theorems, and a KAM reducibility step. It is carefully written and, as far as I can tell, internally coherent.\n\nWhat's new: previous quasi-linear results (Baldi–Berti–Montalto) were small-amplitude only; semilinear KAM for finite-gap tori went back to Kuksin and Kappeler–Pöschel. Here they get arbitrary size. The technical work is substantial: they prove novel extensions of the differential of the canonical map to negative Sobolev spaces (Corollaries 3.3–3.4) and get quantitative Egorov theorems for transport flows. The reduced operator has a leading constant coefficient and the KAM reducibility yields Melnikov non-resonance conditions with the right measure estimates.\n\nThe soft spot is structural rather than a flaw: the entire construction sits on Theorem 3.2, imported from the companion Kappeler–Montalto paper [19]. That theorem gives the canonical chart with pseudo-differential expansion up to remainders of arbitrary negative order, plus tame estimates. If those expansions or the negative-order extensions were off by a derivative, Lemma 6.3 and the rest of Sections 6–7 would not hold. The authors say so themselves—they are not hiding it—and the stress-test confirms no internal contradiction shows up in this manuscript. But this paper is not self-contained: a referee has to treat [19] as part of the proof. That is a heavy but doable review burden.\n\nThe citation pattern looks honest: they cite the semilinear predecessors and the small-amplitude quasi-linear work, and they flag the companion paper clearly. No fitted parameters, no circularity. The claim is what it says.\n\nWho is this for: specialists in KAM theory and Hamiltonian PDEs. It deserves a serious referee, not a desk reject. My recommendation: send it out, but tell the referee to have [19] at hand, and consider a joint referee assignment for both papers. I would likely cite this if I were active in the area.","headline":"First arbitrary-size KAM theorem for quasi-linear KdV perturbations; proof is sound but leans heavily on a companion paper that must be treated as part of the manuscript.","tokens_in":87974,"tokens_out":2496,"would_cite":true,"duration_ms":27774,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37K55","35Q53","37K10"],"pacs":[],"model":"deepseek-v4-flash","headline":"Multi-solitons of any size survive strong KdV perturbations.","keywords":["KdV equation","KAM theory for PDEs","quasi-linear perturbations","finite-gap solutions","periodic multi-solitons","Birkhoff coordinates","Nash-Moser iteration","pseudo-differential operators"],"falsifier":"Evaluate expansion (3.5) for a large-amplitude one- or two-gap KdV potential and check whether the equality $a^{\\Psi\\top}_{-1}=-a^\\Psi_{-1}$ in (3.7) and the negative-Sobolev remainder bounds of Corollaries 3.3-3.4 hold; an explicit failure at this order would block the reduction. A numerical test of the top-order coefficient after the transport-flow conjugation, checking whether it becomes the constant $m_3=-1+O(\\varepsilon)$ as claimed in Lemma 6.5, would likewise settle the reducibility mechanism.","tokens_in":87001,"feed_emoji":"🌊","tokens_out":6924,"duration_ms":64130,"temperature":0.7,"pith_summary":"This paper proves that the periodic multi-soliton solutions of the KdV equation $\\partial_t u = -\\partial_x^3 u + 6u\\partial_x u$ -- solutions of arbitrary size -- survive as linearly stable quasi-periodic tori when the equation is perturbed by any smooth, sufficiently small, quasi-linear Hamiltonian term. Such perturbations are 'strongly nonlinear': they contain the same third-order derivative as the unperturbed equation, so earlier KAM methods that handled only semilinear perturbations did not apply. The result holds for almost every amplitude parameter in the finite-gap family, in the sense that the exceptional set has measure tending to zero with the perturbation strength. The proof combines a Nash-Moser iteration with a multi-step reduction of the linearized operator to a constant-coefficient operator of order three, using pseudo-differential calculus and canonical coordinates adapted to the unperturbed tori.","feed_headline":"Multi-solitons of any size survive strong KdV perturbations","feed_subtitle":"First KAM theorem for arbitrarily large periodic multi-solitons under quasi-linear Hamiltonian perturbations of KdV.","key_machinery":"The machinery centers on the canonical chart $\\Psi_\\nu$ of [19]: a symplectic diffeomorphism $(\\theta,y,w)\\mapsto q$ near a compact family of $S_+$-gap potentials for which $q(\\theta,\\nu+y,0)$ is the finite-gap potential, $w=0$ describes the torus, and expansions (3.5)-(3.6) express the chart and its transpose as $w+\\sum_{k=1}^M a_{-k}\\partial_x^{-k}w$ plus remainders of arbitrary negative order satisfying tame estimates in positive and negative Sobolev spaces. Using that chart, the paper derives a pseudo-differential expansion of the linearized Hamiltonian operator $L_\\omega$, then reduces it in four stages: a quasi-periodic time reparametrization makes the top-order coefficient satisfy a normalization condition; the transport flow of a first-order vector field, together with a quantitative Egorov theorem, eliminates the $(\\phi,x)$-dependence of the top coefficient; two further symplectic conjugations remove the $x$- and $\\phi$-dependence of the first-order coefficient; and a KAM reducibility iteration diagonalizes the remaining order-zero part for most parameters. This almost-invertibility is then inserted into a Nash-Moser scheme to construct the torus embeddings.","core_discovery":"The paper's central claim is Theorem 1.1. For any $f\\in C^\\infty(\\mathbb{T}^1\\times\\mathbb{R}\\times\\mathbb{R})$ and any finite set $S_+$ of excited modes, there is $\\bar s> (|S_+|+1)/2$ and $\\varepsilon_0>0$ such that for $0<\\varepsilon<\\varepsilon_0$ there is a measurable set $\\Xi_\\varepsilon\\subset\\Xi$ with $|\\Xi\\setminus\\Xi_\\varepsilon|\\to 0$, and for each $\\nu\\in\\Xi_\\varepsilon$ a quasi-periodic solution $u_\\varepsilon(\\omega_\\varepsilon(\\nu)t,x;\\nu)$ of $\\partial_t u=\\partial_x\\nabla H_\\varepsilon(u)$ whose Sobolev distance to the finite-gap solution $q(\\cdot,\\cdot;\\nu)$ tends to zero as $\\varepsilon\\to 0$, with frequency vector $\\omega_\\varepsilon(\\nu)\\to -\\omega^{\\rm kdv}(\\nu)$; the torus is linearly stable. The decisive point is that the size of the finite-gap torus is not assumed small and the perturbation is quasi-linear, meaning its Hamiltonian density may depend on $u_x$, so the perturbing vector field contains $\\partial_x^3$ terms of the same order as the unperturbed one.","pith_inferences":["Because the obstruction to the proof is concentrated in the imported normal-form chart, testing that chart numerically for one nontrivial large two-gap potential would give a cheap indicator of the theorem's reach; the KAM and Nash-Moser estimates themselves are self-contained here.","A testable extension would be to repeat the reduction with nonzero mean $c$ or with only $C^{s_*}$ density $f$; the authors note the argument is written for $c=0$ and $f\\in C^\\infty$ just for simplicity, so the same proof should carry the theorem to those cases with only notational changes.","If the companion normal-form chart is established for the defocusing NLS hierarchy, the same four-stage reduction may transfer, suggesting that the main structural requirement is a normal-form coordinate system with pseudo-differential expansion, not the specific KdV dispersion."],"forward_implications":["For any finite set of excited modes and any smooth quasi-linear Hamiltonian perturbation that is small, the perturbed equation admits quasi-periodic invariant tori for a set of amplitude parameters whose complement has measure tending to zero with the perturbation size.","These tori are linearly stable, so the finite-gap KdV solutions are not merely shadowed but persist in a linearly stable sense.","The theorem supplies the first existence result for quasi-periodic solutions of arbitrary size under strongly nonlinear perturbations of an integrable PDE, removing the small-amplitude barrier that limited earlier KAM theorems.","The authors expect the same method to extend to equations in the KdV hierarchy and to defocusing NLS and mKdV; if that expectation is borne out, large finite-gap tori persist in those models as well."],"supporting_citations":[{"why":"Supplies the canonical coordinates near $S_+$-gap potentials with pseudo-differential expansions and negative-Sobolev extensions stated in Theorem 3.2 and Corollaries 3.3-3.4, on which the whole reduction rests.","marker":"[19]"},{"why":"Provides the global Birkhoff coordinates, the finite-gap structure, and the non-degeneracy and asymptotic properties of KdV frequencies used throughout the scheme.","marker":"[20]"},{"why":"Establishes the small-amplitude quasi-linear KdV KAM scheme whose reparametrization, transport-flow, and Egorov steps are adapted here to large tori.","marker":"[3]"},{"why":"Supplies the modulo-tame operator calculus and the KAM reducibility iteration used to diagonalize the order-zero part in Section 7.","marker":"[10]"},{"why":"Provides the earlier large-torus KAM approach for semilinear perturbations of defocusing NLS and the 1-smoothing context that this paper must go beyond for quasi-linear perturbations.","marker":"[9]"},{"why":"Gives the non-degeneracy of the determinant of $\\partial_I \\omega^{\\rm kdv}$, used in Remark 3.10 to invert the frequency map and to choose the parameter set $\\Xi$.","marker":"[11]"},{"why":"Provides the measure estimates for Melnikov non-resonance conditions in the KAM reducibility scheme, adapted in Proposition 8.6.","marker":"[2]"}],"fun_headline_variants":["Arbitrary-size multi-solitons persist under quasi-linear KdV perturbations","Large KAM tori survive quasi-linear KdV perturbations","No size limit: multi-solitons survive quasi-linear KdV perturbations","Quasi-linear perturbations can't destroy large KAM tori"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes from the companion paper [19] that the canonical coordinates near any compact family of $S_+$-gap potentials satisfy the pseudo-differential expansions and tame estimates of Theorem 3.2, Corollaries 3.3 and 3.4; if those expansions or their negative-Sobolev extensions fail, the reduction of the linearized operator and the Nash-Moser construction collapse.","fun_headline_variants_meta":{"raw":{"variants":["Arbitrary-size multi-solitons persist under quasi-linear KdV perturbations","Large KAM tori survive quasi-linear KdV perturbations","No size limit: multi-solitons survive quasi-linear KdV perturbations","Quasi-linear perturbations can't destroy large KAM tori"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000795,"raw_usage":{"total_tokens":3455,"prompt_tokens":852,"completion_tokens":2603,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":468,"completion_tokens_details":{"reasoning_tokens":2528}},"tokens_in":468,"tokens_out":2603,"duration_ms":18358,"temperature":1.0,"reasoning_tokens":2528,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:29:54.725186+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate expansion (3.5) for a large-amplitude one- or two-gap KdV potential and check whether the equality $a^{\\Psi\\top}_{-1}=-a^\\Psi_{-1}$ in (3.7) and the negative-Sobolev remainder bounds of Corollaries 3.3-3.4 hold; an explicit failure at this order would block the reduction. A numerical test of the top-order coefficient after the transport-flow conjugation, checking whether it becomes the constant $m_3=-1+O(\\varepsilon)$ as claimed in Lemma 6.5, would likewise settle the reducibility mechanism.","supporting_citations":[{"cited_title":"Kappeler, R","cited_arxiv_id":null,"evidence_quote":"Supplies the canonical coordinates near $S_+$-gap potentials with pseudo-differential expansions and negative-Sobolev extensions stated in Theorem 3.2 and Corollaries 3.3-3.4, on which the whole reduction rests."},{"cited_title":"Kappeler, J","cited_arxiv_id":null,"evidence_quote":"Provides the global Birkhoff coordinates, the finite-gap structure, and the non-degeneracy and asymptotic properties of KdV frequencies used throughout the scheme."},{"cited_title":"Baldi, M","cited_arxiv_id":null,"evidence_quote":"Establishes the small-amplitude quasi-linear KdV KAM scheme whose reparametrization, transport-flow, and Egorov steps are adapted here to large tori."},{"cited_title":"Berti, R","cited_arxiv_id":null,"evidence_quote":"Supplies the modulo-tame operator calculus and the KAM reducibility iteration used to diagonalize the order-zero part in Section 7."},{"cited_title":"Berti, T","cited_arxiv_id":null,"evidence_quote":"Provides the earlier large-torus KAM approach for semilinear perturbations of defocusing NLS and the 1-smoothing context that this paper must go beyond for quasi-linear perturbations."},{"cited_title":"Bikbaev, S","cited_arxiv_id":null,"evidence_quote":"Gives the non-degeneracy of the determinant of $\\partial_I \\omega^{\\rm kdv}$, used in Remark 3.10 to invert the frequency map and to choose the parameter set $\\Xi$."},{"cited_title":"Baldi, M","cited_arxiv_id":null,"evidence_quote":"Provides the measure estimates for Melnikov non-resonance conditions in the KAM reducibility scheme, adapted in Proposition 8.6."}],"review_version":1}