REVIEW 3 major objections 6 minor 37 references
Splitting methods for Intermediate Long Wave and perturbed Benjamin--Ono models
T0 review · 3 major / 6 minor · reviewed 2026-08-08 · deepseek-v4-flash
Pith's one-line read The paper proves first-order $H^s$ convergence for a perturbative splitting scheme for the Benjamin–Ono class, requiring only one additional derivative on the initial data.
desk verdict Novel BO-based splitting with first-order H^s convergence at s+1 regularity; the stated theorem has a sign condition typo and the key stability hinge uses a uniformity assumption borrowed from cited work rather than proved. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The machine is the Birkhoff map $\Phi: H^s_0 \to h^{s+1/2}$, a nonlinear Fourier transform in which the unperturbed BO flow is diagonal: $\frac{d}{dt}\zeta_n(v(t))=i\omega_n(v(t))\zeta_n(v(t))$ with $\omega_n(v)=n^2-2\sum_{k\ge 0}\min\{k,n\}|\zeta_k(v)|^2$. Here $h^{s+1/2}$ is the weighted $\ell^2$ sequence space with norm $(\sum_{n\ge1} n^{2s+1}|z_n|^2)^{1/2}$. In these coordinates the splitting map $w\mapsto \varphi^\tau_{\mathrm{BO}} e^{\tau A}w$ is shown, via the analytic derivative bounds (10)–(11), to be Lipschitz with the precise constant $1+c\tau$ on bounded sets; this exact leading constant $1$, rather than some $C>1$, is what prevents the $n=T/\tau$ iterates from compounding an $O(1)$ factor into an exponential-in-$1/\tau$ blow-up. The explicit solution formula (4) supplies $\varphi^\tau_{\mathrm{BO}}$ exactly in time, so the only time error is the split between BO and $A$. Local error is then $O(\|a\|_{\ell^\infty}\tau^2)$ in $H^s$, global stability is carried in $h^{s+1/2}$, and an induction keeps all numerical iterates inside a fixed bounded ball $R_\Phi$.
What would settle it
For the Smith equation (or KdV–BO) at fixed high spatial resolution, take $u_0\in H^{s+1}_0$ with a bounded $H^{s+1}$ norm but provably not in $H^{s+2}$, and measure $\sup_{n\tau\le T}\|u(n\tau)-u^n\|_{H^s}$ for $\tau=2^{-m}$; if the log-log slope is strictly below 1, or the prefactor grows without bound as the $H^{s+1}$ norm stays bounded, Theorem 1.2 is false. A second check: with a multiplier $A$ whose symbol $a(k)$ is bounded but non-decaying, the threshold $\tau_0$ in (18) shrinks like $1/(C_{\mathrm{loc}}\|a\|_{\ell^\infty} e^{\tilde c T}T)$; a counterexample would be a choice for which no positive $\tau_0$ yields the stated bound.
Extended reading notes
Core claim
The central claim is Theorem 1.2: for $s\ge 0$, $T>0$, initial data $u_0\in H^{s+1}_0$, and a Fourier multiplier $A$ with bounded, real, even symbol $a$, if $u$ is the exact solution of (1) and $u^n$ is the splitting scheme $u^{n+1}=\varphi^\tau_{\mathrm{BO}} e^{\tau A}u^n$, then for all sufficiently small $\tau$, $\sup_{n\tau\le T}\|u(n\tau)-u^n\|_{H^s} \le C\|a\|_{\ell^\infty}\tau$, with $C$ depending on $s$, $\|u_0\|_{H^{s+1}}$, $\|a\|_{\ell^\infty}$, and $T$. This is a first-order error bound that asks for only one additional derivative, rather than the three or more required by Lie-splitting analyses for KdV or BO. The paper also proves Theorem 1.3: for the ILW equation with depth $\delta$, the splitting scheme converges to the BO solution with error $(C_0 T e^{C_0 T}+C\tau)e^{-2\delta}$, giving a discrete deep-water limit. Numerically, the scheme keeps the ILW energy nearly conserved up to $T=5000$ under a linear time-step condition and produces soliton-resolution dynamics for ILW and a KdV–BO model.
Load-bearing premise
The load-bearing premise is the imported Proposition 3.1—that the Birkhoff map and its inverse are analytic on bounded subsets of $H^s_0$ with derivative bounds (10)–(11)—together with the assumed existence of a global $H^{s+1}$ solution; if either fails uniformly on a fixed bounded ball, the induction that keeps all iterates in a ball does not close.
Editorial extensions
If this is right
- For initial data in $H^{s+1}_0$, the bound $\sup_{n\tau\le T}\|u(n\tau)-u^n\|_{H^s}\le C\|a\|_{\ell^\infty}\tau$ gives rigorous first-order time convergence with no loss of derivatives beyond the single extra derivative, for ILW, Smith, and KdV–BO equations.
- Because both subflows are evaluated exactly in time, the time-discretization error is purely from splitting; the remaining spatial error is controlled by the spectral schemes built on formula (4), giving stable low-regularity computations.
- For ILW with large depth $\delta$, the scheme tracks the BO solution with exponentially small error $e^{-2\delta}$ plus the $O(\tau)$ splitting error, so a single code covers the ILW-to-BO transition.
- The experiments show energy near-preservation up to $T=5000$ under a linear $\tau\sim K^{-1}$ step, avoiding the quadratic $\tau\sim K^{-2}$ restriction imposed on Lie splitting and RK4.
- The observed dynamics for ILW and KdV–BO are consistent with soliton resolution; for KdV–BO with rational initial data of mass $c$, the observed number of asymptotic solitons is $\lceil 2c/3\rceil$.
Reading between the lines
- The paper leaves a fully discrete convergence analysis in Birkhoff coordinates open (Remark 3.9); a natural next step would be to prove that truncation to $K$ Fourier modes perturbs the first $K$ Birkhoff coordinates in a controllable way, turning the semi-discrete theorem into a fully discrete one.
- The numerical energy near-preservation without a quadratic CFL suggests a modified-energy or adiabatic-invariant explanation in Birkhoff coordinates; the authors note the rigorous justification is ongoing, so this is an inference from their numerics rather than a proved claim.
- The conjectured soliton count $\lceil 2c/3\rceil$ for KdV–BO, if confirmed, would mean a non-integrable perturbation of BO still has a robust mass-based soliton counting law; testing it for other perturbations, such as the Smith equation, would reveal how much of BO's integrable rigidity survives.
- On the full line, the exponential-in-$T$ constant could perhaps be improved to a uniform-in-time bound using dispersion and scattering, as has been done for NLS splitting methods; the paper only remarks on this possibility, so such an extension remains open.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes a first-order time-splitting method for a class of perturbed Benjamin--Ono equations ∂t u = ∂x|D|u − 2u ∂xu + A u, where A is a bounded Fourier multiplier. The scheme composes the exact Benjamin--Ono flow φ^τ_BO (evaluated via Gérard's explicit formula) with the exact linear flow e^{τA}. The main analytic result (Theorem 1.2) is first-order convergence in H^s for initial data u0 ∈ H^{s+1}_0, with a constant linear in ||a||_{ℓ∞}. Theorem 1.3 provides an O(e^{-2δ}) convergence rate for the ILW scheme to the BO flow in the deep-water limit. The paper also reports numerical experiments on energy preservation and long-time dynamics, including a numerical study of soliton resolution for ILW and KdV–BO. The proof strategy is to lift the discrete flow to Birkhoff coordinates for BO and to perform the whole stability and induction argument on the h^{s+1/2} side.
Significance. If the gaps identified below are fixed, the paper would be a significant contribution: it would give a first-order H^s convergence theorem for a quasilinear class of dispersive PDEs with only one additional derivative of regularity, substantially improving on classical Lie splitting analyses for KdV and BO. The use of Birkhoff coordinates to obtain a stability bound with Lipschitz constant 1 + O(τ) is conceptually novel and potentially influential. The deep-water limit result and the numerical evidence for long-time energy preservation without a quadratic time-step restriction are valuable, although the fully discrete analysis is explicitly deferred. The paper clearly credits its reliance on the authors' earlier explicit-formula implementations [3,4], which are used only in the numerics and do not enter the semi-discrete convergence proof.
major comments (3)
- [Section 1, Eq. (2)] Condition (2) states a(−k) = a(k), but for A to map real-valued functions to real-valued functions the correct condition is a(−k) = \overline{a(k)}. All three examples satisfy only the latter: for ILW, a(−k) = −a(k) = \overline{a(k)} since a(k) is purely imaginary; for Smith and KdVBO the same identity holds. As written, Theorem 1.2 does not apply to the examples in the introduction, and Theorem 1.3 inherits the problem. The condition and all subsequent uses must be corrected.
- [Section 3.1, Proposition 3.1 and Lemma 3.3] The local estimates (10)–(11) are pointwise in v: for every v there is a neighborhood V and a constant C. Lemma 3.3 applies these estimates to all w(r) and z(r) along trajectories starting from arbitrary f,g in a ball of radius M, and silently uses a single constant C(M) uniformly over the entire ball. The sentence 'Thanks to (10), for any ball B of H^s_0 there exist two positive constants c, C ...' asserts a uniform norm equivalence that is not a consequence of the stated local derivative bounds. The same uniformity is needed in (15) and throughout the induction in Section 3.4, and in the conversion from h^{s+1/2} estimates back to H^s in Lemma 3.7. Without a global or uniform-on-bounded-sets version of these estimates, the Gronwall argument in Lemma 3.3 and the induction do not close. The authors should either state and prove (or precisely quote from [24]) that Φ and Φ^{-1} have derivatives uniformly bounded on bounded subsets of H^s_0, or restructure the stability argument to avoid this uniformity requirement.
- [Section 3.5, proof of Theorem 1.3] The proof applies Theorem 1.2 to the ILW flow uδ, which requires a bound on sup_{t∈[0,T]} ||uδ(t)||_{H^{s+1}} that is uniform in δ for all δ ≥ δ0. However, Proposition 3.10 only provides uniform boundedness in H^s. If the H^{s+1} norm of uδ can grow with δ (or if no uniform bound is known), the constant C in Theorem 1.2 depends on δ, and the stated e^{-2δ} rate does not follow. The authors should supply a uniform H^{s+1} bound — either by proof or by precise citation — or else state Theorem 1.3 with a constant that may depend on the depth parameter in a controlled way.
minor comments (6)
- [Section 3.4, after Eq. (15)] The reference 'By Lemma 3.1' should be 'By Proposition 3.1'; there is no Lemma 3.1 in the manuscript.
- [Section 4, Eq. (20)] The expression u^0_K = (Π_K + Π_K) u0 is a typo; it should involve the conjugate projector, e.g., Π_K + \overline{Π}_K, to define a real-valued initial condition.
- [Section 4.1, initial data construction] The symmetry condition for real initial data should be \hat{U}_{−k} = \overline{\hat{U}_k}, not \hat{U}_{−k} = \hat{U}_k as stated.
- [Introduction, after Eq. (1)] The sentence 'global well-posedness for u0 ∈ H^s_0, s ≥ 0, has recently been obtained in Gassot–Laurens [20]' attributes to [20] a statement about the whole class (1), but [20] treats the ILW equation. Theorem 1.2 is conditional on an existing solution in C([0,T], H^{s+1}_0), so this does not invalidate the theorem, but the presentation should be corrected to avoid overstating the well-posedness theory for the non-integrable examples.
- [Section 1, KdV–BO equation] The sentence 'which can we written in the form (1)' contains a typo ('can we written' should be 'can be written').
- [Remark 1.5] The Galilean transformation formula contains the term −1/a(0), which is undefined when a(0)=0; the formula should be stated in a way that covers the examples.
Circularity Check
No significant circularity: the convergence theorem is derived from external Birkhoff-map theory and the explicit BO formula; the authors' self-citations are numerical tools only.
full rationale
Neither Theorem 1.2 nor Theorem 1.3 is circular. The scheme (5) is defined by composing the exact BO flow (from Gérard's explicit formula (4)) with the exact linear flow e^{τA}; the proof then derives convergence using standard local-error plus stability estimates (Lemmas 3.3 and 3.5–3.7), the Duhamel formula (12), and Birkhoff-coordinate bounds imported from external works [20,23,24]; none of those inputs is equivalent to the theorem's conclusion. The self-citations [3,4] are used only in Section 4 to approximate φ_BO in fully discrete simulations and in remarks about implementation; they are not load-bearing in the semi-discrete convergence proof, which assumes φ_BO exactly. Acknowledged limitations are explicit and non-circular: Remark 3.9 defers the fully discrete analysis to future work, Remark 4.3 says rigorous energy-preservation is 'subject of ongoing work', and Section 4.2 labels the soliton-count formula as a belief supported by numerics with rigorous justification open. The only substantive concern—that Proposition 3.1's pointwise derivative bounds (10)–(11) are silently converted into uniform ball estimates in Lemma 3.3 and Section 3.4—is a proof-rigor and correctness issue rather than a circular reduction, because Proposition 3.1 is an external analytic fact and is not the same statement as the convergence result being proved.
Assumptions & free parameters
assumptions (8)
- domain assumption The Birkhoff map Φ is an analytic diffeomorphism from bounded subsets of H^s_0 onto bounded subsets of h^{s+1/2}, with derivative bounds (10)-(11).
- domain assumption The Birkhoff coordinates of the BO solution evolve by d/dt ζ_n = i ω_n ζ_n with ω_n = n^2 - 2 Σ min{k,n}|ζ_k|^2.
- domain assumption Gérard's explicit formula (4) gives the exact BO solution at any time t.
- domain assumption The perturbed initial-value problem (1) is globally well-posed in H^s_0 for s ≥ 0 with u0 ∈ H^s_0.
- domain assumption The ILW solutions satisfy the uniform a priori bound of Proposition 3.10 uniformly in δ ≥ δ0.
- standard math H^r(T) for r = s+1 ≥ 1 is an algebra and satisfies the product estimate ∥fg∥_{H^r} ≤ C∥f∥_{H^r}∥g∥_{H^r}.
- standard math The multipliers ∂_x|D| and A commute whenever they are both diagonal Fourier multipliers.
- domain assumption The BO equation is globally well-posed in H^s_0 for s ≥ 0 and φ^t_BO maps bounded sets to bounded sets.
Cite this review
Pith. "Pith review of Splitting methods for Intermediate Long Wave and perturbed Benjamin--Ono models." pith.science (2026). https://pith.science/paper/3FT3W6EO
@misc{pith2026260805540,
author = {Pith},
title = {Pith review of: Splitting methods for Intermediate Long Wave and perturbed Benjamin--Ono models},
year = {2026},
howpublished = {\url{https://pith.science/paper/3FT3W6EO}},
note = {Machine review of arXiv:2608.05540}
}
abstract
Recently, novel integrability techniques, and most notably Birkhoff coordinates followed by an explicit formula, have been introduced to solve the Benjamin--Ono (BO) equation, leading to major theoretical and computational advances. We propose to perturb this explicit BO formula via a novel splitting method for numerically solving a class of PDEs, which includes the Intermediate Long Wave (ILW) equation, as well as other quasilinear equations that are not necessarily integrable. Using the Birkhoff coordinates of BO we establish first-order convergence in the $H^s$-norm ($s \ge 0$) while requiring only one additional derivative on the initial data $u_0 \in H^{s+1}_{0}$, significantly improving upon the regularity requirements of typical splitting methods for nonlinear dispersive equations. Furthermore, we prove that in the deep-water limit, the scheme for the ILW equation converges to the BO solution. Computational advantages of these schemes are shown in simulations: unlike classical splitting methods, our approach does not require a restrictive quadratic time step condition to nearly preserve energy over long time scales, thus rendering efficient and accurate long-time simulations feasible. As an application, we numerically explore the soliton resolution conjecture for both the ILW and a KdV--BO equation.
Figures
Figures from the paper (3 more)
Reference graph
Works this paper leans on
- [24]
-
[1]
Y. Alama Bronsard,A symmetric low-regularity integrator for the nonlinear Schrödinger equation, IMA Journal of Numerical Analysis,44:3648–3682, (2024)
work page 2024
-
[2]
Y. Alama Bronsard, Y. Bruned, G. Maierhofer and K. Schratz,Symmetric resonance based integrators and forest formulae, Found. Comput. Math., (2026)
work page 2026
-
[3]
Y. Alama Bronsard, X. Chen and M. Dolbeault,Spectrally accurate fully discrete schemes for some non- local and nonlinear integrable PDEs via explicit formulas. Foundations of Computational Mathematics, 1–28 (2026)
work page 2026
-
[4]
Y. Alama Bronsard and T. Laurens,On the convergence of explicit formulas forL2 solutions to the Benjamin-Ono and continuum Calogero-Moser equations, arXiv preprint arXiv:2602.19046, (2026)
arXiv 2026
-
[5]
Y. Alama Bronsard and O. Mæhlen,Large-torus limits of explicit formulas and applications to long-time dynamics of the Benjamin–Ono equation onR, work in preparation, (2026)
work page 2026
-
[6]
D. Bambusi and P. Gérard,A Nekhoroshev theorem for some perturbations of the Benjamin-Ono equa- tion with initial data close to finite gap tori, Math. Z.307(3):54, (2024)
work page 2024
-
[7]
J. Bernier and B. Grébert,Long time dynamics for generalized Korteweg-de Vries and Benjamin-Ono equations, Arch. Ration. Mech. Anal.241(3):1139–1241, (2021)
work page 2021
Show all 37 references
-
[8]
Carles and C
R. Carles and C. Su.Scattering and uniform in time error estimates for splitting method in NLS, Found. Comput. Math.24:683–722, (2024)
2024
-
[9]
Chapouto, G
A. Chapouto, G. Li, T. Oh and D. Pilod,Deep-water limit of the intermediate long wave equation in L2, Math. Res. Lett.31(6):1655–1692, (2024)
2024
-
[10]
Chapouto, J
A. Chapouto, J. Forlano, G. Li, T. Oh and D. Pilod,Intermediate long wave equation in negative Sobolev spaces, Proc. Amer. Math. Soc. Ser. B,11:452–468, (2024)
2024
-
[11]
Cohen, E
D. Cohen, E. Hairer and C. Lubich,Conservation of energy, momentum and actions in numerical discretizations of non-linear wave equations, Numer. Math.,110(2):113–143, (2008)
2008
-
[12]
Deng and H
Z. Deng and H. Ma,Optimal error estimates of the Fourier spectral method for a class of nonlocal, nonlinear dispersive wave equations. Appl. Numer. Math.59:988–1010, (2009)
2009
-
[13]
Dutta, H
R. Dutta, H. Holden, U. Koley, and N. H. Risebro,Operator splitting for the Benjamin–Ono equation, J. Differential Equations259(2015), no. 11, 6694–6717
2015
-
[14]
Faou,Geometric numerical integration and Schrödinger equations, vol
E. Faou,Geometric numerical integration and Schrödinger equations, vol. 15, European Mathematical Society, (2012)
2012
-
[15]
Faou and B
E. Faou and B. Grébert,Hamiltonian interpolation of splitting approximations for nonlinear PDEs, Found. Comput. Math.,11(4):381–415, (2011)
2011
-
[16]
E. Faou, B. Grébert and E. Paturel,Birkhoff normal form for splitting methods applied to semilinear Hamiltonian PDEs. I. Finite-dimensional discretization, Numer. Math.,114(3):429–458, (2010). 28
2010
-
[17]
E. Faou, B. Grébert and E. Paturel,Birkhoff normal form for splitting methods applied to semilinear Hamiltonian PDEs. II. Abstract splitting, Numer. Math.,114(3):459–490, (2010)
2010
-
[18]
Y. Feng, G. Maierhofer and C. Wang,Explicit symmetric low-regularity integrators for the nonlinear Schrödinger equation, SIAM J. Sci. Comput.,47(4):2154, (2025)
2025
-
[19]
Gassot, P
L. Gassot, P. Gérard and P. D. Miller,A proof of the soliton resolution conjecture for the Benjamin–Ono equation, arXiv preprint arXiv:2601.10488, (2026)
2026 arXiv
-
[20]
Gassot and T
L. Gassot and T. Laurens,Global well-posedness for the ILW equation inHs(T)fors >−1/2, arXiv preprint arXiv:2506.05149, (2025)
2025 arXiv
-
[21]
Gauckler and C
L. Gauckler and C. Lubich,Splitting integrators for nonlinear Schrödinger equations over long times, Found. Comput. Math.,10(3):275–302, (2010)
2010
-
[22]
Gérard,An explicit formula for the Benjamin–Ono equation, Tunisian J
P. Gérard,An explicit formula for the Benjamin–Ono equation, Tunisian J. Math.,5(3): 593-603, (2023)
2023
-
[23]
Gérard and T
P. Gérard and T. Kappeler,On the integrability of the Benjamin–Ono equation on the torus, Comm. Pure Appl. Math.,74:1685–1747 (2021)
2021
-
[25]
Gérard, T
P. Gérard, T. Kappeler, and P. Topalov,Sharp well-posedness results of the Benjamin–Ono equation in H s(T,R)and qualitative properties of its solutions, Acta Math.231(2023), no. 1, 31–88
2023
-
[26]
Harrop-Griffiths, R
B. Harrop-Griffiths, R. Killip, and M. Vişan,A priori bounds and equicontinuity of orbits for the inter- mediate long wave equation, J. Evol. Equ.,26(2026)
2026
-
[27]
Holden, K
H. Holden, K. Karlsen, N. Risebro and T. Tao,Operator splitting for the KdV equation. Math. Comp., 80(274): 821-846 (2011)
2011
-
[28]
Holden, C
H. Holden, C. Lubich and N. Risebro,Operator splitting for partial differential equations with Burgers nonlinearity. Math. Comp.,82(281): 173-185 (2013)
2013
-
[29]
Ifrim and J.-C
M. Ifrim and J.-C. Saut,The lifespan of small data solutions for Intermediate Long Wave equation (ILW). Commun. Partial Differ. Equ.,50(3): 258-300 (2025)
2025
-
[30]
Klein and J.-C
C. Klein and J.-C. Saut,Nonlinear dispersive equations–Inverse Scattering and PDE methods, Appl. Math. Sci. 209, Springer International Publishing, (2021)
2021
-
[31]
Liu and X
J. Liu and X. Yuan,A KAM theorem for Hamiltonian partial differential equations with unbounded perturbations, Comm. Math. Phys.307(3):629, (2011)
2011
-
[32]
Marissaël,A scalar asymptotic model for long waves with KdV and Shrödinger behaviors, work in preparation, (2026)
A. Marissaël,A scalar asymptotic model for long waves with KdV and Shrödinger behaviors, work in preparation, (2026)
2026
-
[33]
Mi and K
L. Mi and K. Zhang,Invariant tori for Benjamin-Ono equation with unbounded quasi-periodically forced perturbation, Discrete Contin. Dyn. Syst. A34(2):689–707, (2014)
2014
-
[34]
Paulsen,Justification of the Benjamin–Ono equation as an internal water waves model, Ann
M.O. Paulsen,Justification of the Benjamin–Ono equation as an internal water waves model, Ann. PDE. 10(25): 1-129, (2024)
2024
-
[35]
Pelloni and B., V.A
B. Pelloni and B., V.A. Dougalis,Error estimate for a fully discrete spectral scheme for a class of nonlinear, nonlocal dispersive wave equations. Appl. Numer. Math.37:95–107, (2001)
2001
-
[36]
Rousset and K
F. Rousset and K. Schratz,Convergence error estimates at low regularity for time discretizations of KdV, Pure Appl. Anal.,4(1):127–152, (2022)
2022
-
[37]
Smith,Nonlinear Kelvin and continental-shelf waves, Journal of Fluid Mechanics.52(2):379-391, (1972)
R. Smith,Nonlinear Kelvin and continental-shelf waves, Journal of Fluid Mechanics.52(2):379-391, (1972)
1972
Reviewed August 8, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.