{"id":"e9f9c438-a8c8-4edc-9883-52c7432ada01","arxiv_id":"2412.17117","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"KdVH has additional solitary and periodic traveling waves beyond KdV solitons, and the paper's ImEx-SBP schemes provably preserve energy and are asymptotic preserving toward KdV.","lead":"This paper studies the hyperbolized KdV equation KdVH, finding new traveling wave families and designing time and space discretizations that preserve energy and converge to the original KdV equation as a relaxation parameter shrinks. It matters because it supplies validated, structure-preserving numerical tools for a water-wave model whose dynamics are richer, and messier, than KdV's.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"AP theorems are proven only at the level of a formal Hilbert expansion; the missing validity/stability argument is the main load-bearing soft spot.","rationale":"The paper is a solid numerical analysis contribution. The AP argument is standard in style, the energy-conservation proofs and SBP constructions check out algebraically, and the numerical experiments are detailed with a reproducibility repository. The reader's weakest-assumption choice (well-prepared data for type II methods) is correct as far as it goes; I agree that without well-preparedness the auxiliary components are not AP, as the paper itself acknowledges. My concern is slightly broader: even the type I result for u rests on a formal Hilbert expansion and does not supply the stability half of Definition 1. This is a proof-technical gap rather than a demonstrated counterexample, and the tables are consistent with the formal conclusions. I also noted the Figure 6 caption misstates SSP3-ImEx(3,4,3) as AA for all components when the text and tables say only u is guaranteed; this is presentation-level and does not affect the central claims. Overall the evidence supports the verdict, so I recommend no change.","tokens_in":30279,"tokens_out":20400,"duration_ms":183689,"concrete_test":"Run the Table 1 experiment (SSP2-ImEx(2,2,2), Delta t = 0.005, SBP FD with N = 2^10, domain [-40,40], final time 16.67) with the same well-prepared u0 but bounded non-well-prepared v0 = 0 and w0 = 0, for tau = 10^-1, 10^-3, 10^-5, 10^-7, 10^-9. If ||u - eta_Delta||_M still tends to zero (ideally ~O(tau)) while ||v - D_- eta|| and ||w - DD_- eta|| remain O(1), then Theorem 1's 'always AP for u' is supported; if ||u - eta_Delta|| plateaus above zero, the formal-expansion proof has a real hole.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorems 1 and 2 prove asymptotic preservation by inserting a Hilbert expansion (23) for the numerical solution and stage vectors and by matching powers of tau. No argument establishes that this expansion is valid for the actual fully discrete solution, and for stiff relaxation systems with non-well-prepared data one generally expects initial-layer contributions that are not captured by a power series. The theorems also conclude un+1 - eta(tn+1) = O(Delta t^p) without proving stability of the limiting ImEx method for KdV, although Definition 1 makes stability part of the AP property. Thus the proofs establish consistency of the formal limit rather than the full AP property. The paper's own Remark 4 and Tables 1-2 show that formally omitted terms can matter for v and w; the analogous possibility for u is the residual risk. This is the weakest load-bearing spot because the headline numerical claim depends on tau -> 0 convergence at fixed Delta t, not only on the algebraic limit of the Butcher tableau.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies a hyperbolic approximation of the Korteweg-de Vries equation (KdVH). In Section 2, the authors analyze traveling-wave solutions, showing that KdV-soliton-like solutions exist up to a maximum speed and numerically identifying other solitary and periodic waves, including peaked and left-going waves related to Camassa-Holm/Degasperis-Procesi models. In Section 3, they propose an ImEx-RK splitting (15) in which only the nonlinear convective term is explicit and all linear-derivative and relaxation terms are implicit. They prove asymptotic-preserving (AP) properties for type I and type II ImEx methods (Theorems 1 and 2), develop SBP-based energy-conserving semidiscretizations (Theorems 3 and 4), and use entropy relaxation to obtain fully discrete energy conservation. Section 4 provides extensive numerical evidence for the AP and energy-conservation properties, using finite difference, DG, and Fourier spatial discretizations. A reproducibility repository is provided.","tokens_in":30456,"tokens_out":9422,"duration_ms":89497,"significance":"The paper is valuable: it offers a practical route to solving KdV via a hyperbolic system while preserving energy, and it extends the AP methodology from classical relaxation systems to a third-order dispersive system. The energy-conservation proofs (Theorems 3 and 4) are clean, and the numerical experiments are thorough, comparing four type I and four type II methods and confirming the role of the globally-stiffly-accurate (GSA) property in the AP behavior of auxiliary variables. The connection to Camassa-Holm-type traveling waves is interesting and well illustrated. If the AP theorems are made rigorous, the methods would be a significant contribution to structure-preserving discretizations for hyperbolized dispersive PDEs. The reproducibility repository and the explicit discussion of well-prepared initial data are also strengths.","major_comments":[{"comment":"The AP property defined in Definition 1 requires that the limiting discretization P^0_h be consistent and stable. However, the proofs of Theorems 1 and 2 assume the Hilbert expansions (23) for the numerical solution and stage vectors without justification, and they only verify that the leading-order terms satisfy the same algebraic scheme as the ImEx-RK method applied to KdV. No stability of the limiting scheme is established, and no argument is given that the expansions represent the actual discrete solution uniformly in tau for fixed Delta t. Since the theorem statements (19)-(20) are unconditional, they are stronger than what is proven; at present they establish consistency of the formal limit rather than the full AP property. Please either provide a validity/stability argument for the expansions or restate the results as formal AP statements with the Hilbert-expansion assumption included in the hypotheses and with the stability part of Definition 1 handled separately.","section":"Section 3.1 (Theorems 1 and 2)"},{"comment":"In the proof of Theorem 2, the propagation of well-preparedness from time t_n to t_{n+1} is justified via the GSA property, but the key identity 'vn+1 = b^T A^{-1} vv = v(s)' is only sketched. For type II methods, the update (33c) involves the first stage q^(1)=q^n and the reduced matrices with hats; a complete derivation of the final-stage identity for the auxiliary components would remove ambiguity and make the induction argument explicit.","section":"Section 3.1, proof of Theorem 2"}],"minor_comments":[{"comment":"In equation (59), the term 'wwwT DT+Muuu' appears to be a typesetting error; it should presumably read 'www^T M D_+ uuu'. The argument is correct, but the notation is confusing as printed.","section":"Section 3.2, proof of Theorem 4"},{"comment":"The inner product in formula (61) is not defined. Since the modified energy (47) is a weighted L2 norm with weights (1, tau, tau), and its discrete counterpart involves the mass matrix M, please state explicitly that the inner product in (61) is the corresponding weighted discrete inner product; otherwise the formula appears to conserve the standard Euclidean norm, which would not preserve I(q).","section":"Section 3.3, equation (61)"},{"comment":"The sentence 'the system qt + gx = 0 is strictly hyperbolic' is unclear, since g is not a flux function but a differential operator. Please rephrase, for example by stating that the linear part of the system is hyperbolic in the sense of the eigenvalue analysis in Remark 2.","section":"Section 3.1, after equation (15)"},{"comment":"The claim that the linearization of (11) is ill-posed for c tau > 0 is stated without derivation; a one-line justification (e.g., the dispersion relation of the linearized equation) would help the reader.","section":"Section 2.2, Remark 1"},{"comment":"The statement that KdVH imposes a maximum speed on soliton-like waves is supported by the linear analysis at u=0 in Remark 2, but the precise condition c^2 < 1/tau under which the homoclinic orbit exists is derived in Section 2.1. Please connect these two statements explicitly.","section":"Section 2.1 and Remark 2"},{"comment":"The estimated orders of convergence would be more informative if the fixed time step Delta t = 0.005 were stated in the table captions rather than only in the preceding text.","section":"Section 4.1, Tables 1-9"}],"recommendation":"major_revision","confidential_remarks":"The paper is well written and the reproducibility of the numerical experiments is a clear strength. The main concern is that the AP theorems are formal: they assume Hilbert expansions without a validity argument and do not address stability of the limiting scheme, despite Definition 1 requiring stability. I believe this is fixable by either adding a rigorous stability/validity discussion or restating the results as formal AP properties with the assumptions made explicit in the hypotheses. After such a revision, the paper would be close to acceptable; I do not see a fundamental flaw or a scope mismatch."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Honestly, this is a good paper and the reader's ACCEPT is about right. The AP proofs have a formal-expansion gap that deserves flagging, but it is a gap in the proof of the advertised claim, not a sign the claim is false; the numerics support it.\n\nWhat's new: the traveling wave analysis is the most interesting part. They show KdVH solitons converge to KdV solitons, and they exhibit new traveling waves (peaked, left-going, large-speed) with no KdV counterpart, plausibly related to Camassa-Holm/Degasperis-Procesi. These are computed with Petviashvili and not proven to exist, but the phase-plane story is convincing. The numerical part: the splitting (15) is nonstandard, and Theorems 1-2 give useful AP conditions for that splitting. The energy-preserving SBP semidiscretization and relaxation RK fully discrete schemes are clean and verified by long-time experiments. Reproducibility repo is a plus.\n\nSoft spots, in proportion. The stress-test note is on target: Theorems 1 and 2 prove AP by inserting Hilbert expansions and matching powers of tau. They never justify that the expansion exists for the actual discrete solution, and the O(Δt^p) estimates assume stability of the limiting ImEx method for KdV. Since Definition 1 requires stability, the theorems overstate what is proven--they establish formal consistency of the limit discretization. This is a real caveat, though for GSA methods the tables show clean rates and the failure modes for non-GSA methods are exactly as expected. The type II theorem also needs well-prepared data; Remark 4's induction only works because GSA forces the update onto the manifold, which is fine but should be stated as a projection argument.\n\nThe 'other traveling waves' are numerically demonstrated, not analytically proven; the paper is honest about that, but a referee should ask for a clear statement that these are numerical observations. Also the caption of Figure 6 lists SSP3-ImEx(3,4,3) as AA for all components, while the text and the figure itself say only u; that's an inconsistency that should be fixed.\n\nBottom line: this deserves a serious referee. I'd send it out, with the AP definition gap as the main request for revision. For a reader working on hyperbolic relaxation or structure-preserving methods, this is a useful paper; I'd bring it to reading group and would cite it.","headline":"Worth refereeing: strong numerical paper with a real but non-fatal gap between the formal AP proofs and the advertised stability claim, plus a rich traveling-wave catalog that is numerically convincing but not proven.","tokens_in":30987,"tokens_out":3083,"would_cite":true,"duration_ms":28223,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q53","65L06","65M06","65M12","65M70"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a specific implicit-explicit Runge-Kutta splitting for the hyperbolic KdV approximation is asymptotic preserving and shows how to conserve its modified energy, while also classifying its traveling-wave solutions.","keywords":["hyperbolic relaxation approximation","Korteweg-de Vries equation","traveling waves","asymptotic-preserving schemes","implicit-explicit Runge-Kutta methods","summation-by-parts operators","energy-preserving discretization","solitons"],"falsifier":"Run a globally stiffly accurate type I method such as AGSA(3,4,2) on the KdVH system with well-prepared data at fixed $\\Delta t$, and measure $\\|v^n - D_-\\eta^n\\|_2$ as $\\tau\\to0$; Theorem 1 predicts this error is $O(\\Delta t^p)$. If the observed error plateau is orders of magnitude larger than $\\Delta t^p$ as $\\tau$ decreases, the theorem's claim that GSA suffices for the auxiliary components is false.","tokens_in":30098,"feed_emoji":"🌊","tokens_out":8822,"duration_ms":78940,"temperature":0.7,"pith_summary":"The paper studies KdVH, a hyperbolic approximation of the Korteweg-de Vries equation that replaces the third derivative with two auxiliary variables governed by relaxation equations. Its main mathematical content is a proof that a carefully chosen implicit-explicit Runge-Kutta splitting is asymptotic preserving: as the relaxation parameter $\\tau$ tends to zero, the numerical solution converges to a consistent discretization of KdV. For type I methods this holds unconditionally for the $u$ component and, for globally stiffly accurate methods, for the auxiliary components $v$ and $w$; type II methods need well-prepared initial data. On the solution side, the paper shows that KdVH supports solitary waves that approach KdV solitons, plus additional periodic and peaked traveling waves reminiscent of other water-wave models. It also constructs spatial semidiscretizations based on summation-by-parts operators and uses entropy relaxation to make the fully discrete schemes energy-preserving.","feed_headline":"One splitting rule makes KdVH solvers asymptotic preserving","feed_subtitle":"With globally stiffly accurate time stepping, the method recovers KdV as the relaxation parameter vanishes and conserves energy.","key_machinery":"The load-bearing object is the splitting $q_t=f(q)+g(q)$ of the KdVH system, with $f(q)=(-u u_x,0,0)^T$ and $g(q)=(-w_x,(v_x-w)/\\tau,(-u_x+v)/\\tau)^T$. The asymptotic-preservation argument works by inserting formal Hilbert expansions in powers of $\\tau$ into the Runge-Kutta stage equations; invertibility of the implicit coefficient matrix $A$ forces $v\\to u_x$ and $w\\to v_x$ at leading order, and the $u$ update collapses to the same implicit-explicit discretization of KdV. The globally stiffly accurate property, meaning the last stage coincides with the step update, is what keeps the auxiliary variables on the equilibrium manifold. For energy conservation, the paper uses periodic summation-by-parts operators satisfying $MD+D^TM=0$ with split-form nonlinear terms, and then inserts an entropy relaxation parameter $\\gamma_n$ into the time update to enforce preservation of the discrete modified energy.","core_discovery":"The central claim is that KdVH is a usable hyperbolic surrogate for KdV: its solitary waves converge to KdV solitons as $\\tau\\to0$, and its dispersion relation has a finite maximum speed that removes the unbounded phase speeds of KdV. The numerical core claim is Theorem 1: for the splitting that treats only the nonlinear convection explicitly and everything else implicitly, an implicit-explicit Runge-Kutta method of type I is always asymptotic preserving for the $u$ component, with $u^{n+1}-\\eta(t_{n+1})=O(\\Delta t^p)$ in the stiff limit; if the method is also globally stiffly accurate, the auxiliary components satisfy $v^{n+1}-\\eta_x(t_{n+1})=O(\\Delta t^p)$ and $w^{n+1}-\\eta_{xx}(t_{n+1})=O(\\Delta t^p)$. A parallel statement holds for globally stiffly accurate type II methods with well-prepared initial data. The paper also proves that upwind summation-by-parts spatial discretizations conserve the discrete mass and modified energy, and that the entropy relaxation technique extends this conservation to the fully discrete level.","pith_inferences":["A consequence the paper leaves implicit is that the finite maximum speed of KdVH soliton-like waves may act as a built-in regularization of KdV's unphysical arbitrarily fast solitons; the paper notes the speed bound but does not develop it as a modelling feature.","The reduction to the dispersive equation (11) suggests that the peaked traveling waves of KdVH could be compared quantitatively with exact solutions of integrable shallow-water models, a comparison not made here.","The same explicit/implicit splitting principle should transfer to other hyperbolized dispersive equations, where the equilibrium relations $v=u_x$, $w=u_{xx}$ are replaced by the corresponding algebraic constraints of the target model."],"forward_implications":["The asymptotic-preserving property means a fixed discretization of KdVH can be run with a small but finite $\\tau$ and still produce a consistent approximation of the KdV equation, so KdVH can be used as a hyperbolic surrogate in codes that need first-order fluxes or nonreflecting boundary conditions.","The globally stiffly accurate condition is not a technicality: without it, the auxiliary variables $v$ and $w$ fail to converge as $\\tau\\to0$, as shown in the paper's tables, so users who need derivatives of the solution must choose GSA methods.","The energy-preserving full discretizations give linear error growth over long times instead of quadratic, matching the known behavior of conservative integrators for KdV itself.","The newly identified peaked and left-going traveling waves are genuine solutions of KdVH but are not KdV solitons, so any simulation using KdVH as a KdV proxy must be aware that these extra solutions exist and may be dynamically stable.","The results transfer from finite difference to discontinuous Galerkin and Fourier spatial discretizations, since the analysis only requires the summation-by-parts structure."],"supporting_citations":[{"why":"Proposes the KdVH system and its dispersion analysis, the object whose solutions and discretizations this paper extends.","marker":"[4]"},{"why":"Supplies the asymptotic-preserving framework, the type I/type II ImEx-RK classification, and the globally stiffly accurate condition adapted in Theorems 1 and 2.","marker":"[9]"},{"why":"Provides the type II additive Runge-Kutta methods used in the numerical AP/AA tests, including cases whose observed behavior exceeds the theoretical guarantees.","marker":"[34]"},{"why":"Introduces the entropy relaxation Runge-Kutta technique used to enforce fully discrete energy conservation.","marker":"[35]"},{"why":"Supplies the diagonal-norm upwind summation-by-parts operators used for the energy-conserving spatial semidiscretizations.","marker":"[40]"},{"why":"Establishes asymptotic-preserving and asymptotically-accurate properties for ImEx-RK methods on hyperbolic relaxation systems, the template for the proofs here.","marker":"[42]"},{"why":"Shows the chain rule has no discrete analogue, motivating the split-form SBP treatment of the nonlinear term.","marker":"[43]"},{"why":"Demonstrates linear versus quadratic error growth for conservative versus nonconservative KdV integrators, the benchmark for the long-time energy experiments.","marker":"[17]"}],"fun_headline_variants":["KdVH solvers stay energy-conserving and asymptotic preserving","A single splitting rule yields asymptotic preservation for KdVH","Energy-conserving and asymptotic-preserving schemes for KdVH","KdVH solitary waves approach KdV solitons"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The asymptotic-preservation results for the auxiliary components assume the initial data are well prepared, meaning $v(0)=u_x(0)+O(\\tau)$ and $w(0)=u_{xx}(0)+O(\\tau)$, and the proof additionally assumes a formal Hilbert expansion in powers of $\\tau$.","fun_headline_variants_meta":{"raw":{"variants":["KdVH solvers stay energy-conserving and asymptotic preserving","A single splitting rule yields asymptotic preservation for KdVH","Energy-conserving and asymptotic-preserving schemes for KdVH","KdVH solitary waves approach KdV solitons"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001388,"raw_usage":{"total_tokens":5616,"prompt_tokens":939,"completion_tokens":4677,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":555,"completion_tokens_details":{"reasoning_tokens":4606}},"tokens_in":555,"tokens_out":4677,"duration_ms":31881,"temperature":1.0,"reasoning_tokens":4606,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T05:47:18.901667+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a globally stiffly accurate type I method such as AGSA(3,4,2) on the KdVH system with well-prepared data at fixed $\\Delta t$, and measure $\\|v^n - D_-\\eta^n\\|_2$ as $\\tau\\to0$; Theorem 1 predicts this error is $O(\\Delta t^p)$. If the observed error plateau is orders of magnitude larger than $\\Delta t^p$ as $\\tau$ decreases, the theorem's claim that GSA suffices for the auxiliary components is false.","supporting_citations":[{"cited_title":"Perfectly matched layers methods for mixed hyperbolic–dispersive equations","cited_arxiv_id":null,"evidence_quote":"Proposes the KdVH system and its dispersion analysis, the object whose solutions and discretizations this paper extends."},{"cited_title":"Asymptotic preserving methods for quasilinear hyperbolic systems with stiff relaxation: a review","cited_arxiv_id":null,"evidence_quote":"Supplies the asymptotic-preserving framework, the type I/type II ImEx-RK classification, and the globally stiffly accurate condition adapted in Theorems 1 and 2."},{"cited_title":"Additive Runge–Kutta schemes for convection– diffusion–reaction equations","cited_arxiv_id":null,"evidence_quote":"Provides the type II additive Runge-Kutta methods used in the numerical AP/AA tests, including cases whose observed behavior exceeds the theoretical guarantees."},{"cited_title":"Relaxation Runge–Kutta methods: Conservation and stability for inner-product norms","cited_arxiv_id":null,"evidence_quote":"Introduces the entropy relaxation Runge-Kutta technique used to enforce fully discrete energy conservation."},{"cited_title":"Diagonal-norm upwind SBP operators","cited_arxiv_id":null,"evidence_quote":"Supplies the diagonal-norm upwind summation-by-parts operators used for the energy-conserving spatial semidiscretizations."},{"cited_title":"Implicit–explicit Runge–Kutta schemes and applications to hyperbolic systems with relaxation","cited_arxiv_id":null,"evidence_quote":"Establishes asymptotic-preserving and asymptotically-accurate properties for ImEx-RK methods on hyperbolic relaxation systems, the template for the proofs here."},{"cited_title":"Mimetic properties of difference operators: Product and chain rules as for func- tions of bounded variation and entropy stability of second derivatives","cited_arxiv_id":null,"evidence_quote":"Shows the chain rule has no discrete analogue, motivating the split-form SBP treatment of the nonlinear term."},{"cited_title":"Accuracy and conservation properties in numerical in- tegration: the case of the Korteweg-de Vries equation","cited_arxiv_id":null,"evidence_quote":"Demonstrates linear versus quadratic error growth for conservative versus nonconservative KdV integrators, the benchmark for the long-time energy experiments."}],"review_version":1}