Error estimates in the non-relativistic limit for the two-dimensional cubic Klein-Gordon equation

Fangzheng Huang; Yong Lu

REVIEW 1 major objections 3 minor 22 references

Error estimates in the non-relativistic limit for the two-dimensional cubic Klein-Gordon equation

T0 review · 1 major / 3 minor · reviewed 2026-08-04 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper proves that the two-dimensional cubic Klein-Gordon equation is approximated by the cubic nonlinear Schrödinger equation with error O(ε²) up to times ε^{-2/(N+1)}.

desk verdict New 2D result that extends the 3D non-relativistic-limit analysis, but it leans on deferred structural lemmas from the authors' prior paper and needs a revision to be fully convincing. read the letter →

arxiv 2509.08271 v1 pith:GVZZF5TZ submitted 2025-09-10 math.AP

Yong Lu , Fangzheng Huang This is my paper · ORCID

classification math.AP MSC 35Q5535L0535B25

keywords non-relativisticlimitcubicKlein-GordonequationnonlinearSchrödingerWKBexpansiongeometricopticserrorestimateslong-timestabilitytwo-dimensional

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper analyzes the small-$\varepsilon$ limit of the two-dimensional cubic Klein-Gordon equation, where $\varepsilon$ is inversely proportional to the speed of light. It claims that in the defocusing case, the solution stays within $O(\varepsilon^2)$ of the cubic nonlinear Schrödinger solution on a long time interval growing like $\varepsilon^{-2/(N+1)}$, with an error prefactor that grows only algebraically in time. With less regular data, the rate slows to $O(\varepsilon)$ on a shorter time interval. This establishes that the Schrödinger approximation is not merely formal but quantitatively accurate over the physically relevant non-relativistic timescale, in Sobolev norms rather than only $L^2$.

What carries the argument

The engine is a WKB expansion with fast phase $\theta = t/\varepsilon^2$: the solution is written as a trigonometric polynomial $\sum_n \varepsilon^n \sum_p e^{ip\theta}U_{n,p}$, inserted into an equivalent symmetric hyperbolic system for $U = (\varepsilon\nabla u, \varepsilon^2\partial_t u, u)$. The lowest-order amplitudes satisfy the cubic NLS, and the higher profiles satisfy the linearized Schrödinger equations (3.47)–(3.48). The crucial mechanism is that the algebraic structure of these profiles cancels all resonant terms, leaving an error $\varepsilon^{K+1}$ that is controlled by Duhamel stability together with the $t^{-1} L^\infty$ decay of the defocusing cubic NLS.

What would settle it

Compute $u$ numerically for the defocusing 2D cubic Klein-Gordon equation with $\varepsilon = 10^{-2}$ and smooth localized data, solving the cubic NLS for $g_0$ simultaneously, and measure $\|u - (e^{i\theta}g_0 + e^{-i\theta}\bar{g}_0)\|_{H^{s-8}}$ up to $t = \varepsilon^{-2/(N+1)}$; the theorem predicts this stays $O((1+t)^N \varepsilon^2)$, so observing $O(\varepsilon)$ growth, loss of $H^{s-8}$ regularity, or breakdown before $T_0 \varepsilon^{-\alpha}$ would refute it. As an analytic check, one can verify $\Phi_{n,p} = 0$ in (2.5) for $n \leq K$ using the explicit recursive $U_{n,p}$; any nonzero residual at order $\leq K$ contradicts the construction.

Watch

Extended reading notes

Core claim

The central claim is that, for $\lambda>0$ and initial data $(\varphi,\psi)$ in $H^s$ with $s>9$ plus weighted regularity, the solution $u$ of $\varepsilon^2\partial_{tt}u - \Delta u + \varepsilon^{-2}u + \lambda u^3 = 0$ satisfies $\|u - (e^{i\theta}g_0 + e^{-i\theta}\bar{g}_0)\|_{H^{s-8}} \leq C_{\varphi,\psi}(1+t)^{N_{\varphi,\psi}} \varepsilon^2$ for $t \leq T_0 \varepsilon^{-2/(N+1)}$, where $\theta = t/\varepsilon^2$ and $g_0$ solves the cubic NLS $2i\partial_t g_0 - \Delta g_0 + 3\lambda|g_0|^2g_0 = 0$ with $g_0(0) = (\varphi - i\psi)/2$. For initial data with limited regularity, $s>5$, the error is $O((1+t) + (1+t)^{\tilde N})\varepsilon$ over a shorter time interval. The proof constructs WKB approximate solutions to arbitrary even order $K$, yielding error $\varepsilon^{K+1}$ in $H^{s-2K-4}$ over the same long-time scale, and for the focusing case the approximation holds until the NLS solution's maximal ex

Load-bearing premise

The paper assumes, without proving here, that the WKB profiles have the exact algebraic form stated in Propositions 3.4, 3.5, and Corollary 3.6 and satisfy the profile equations (3.47)–(3.48); these structural identities are quoted from a prior paper. If they failed, the approximate solution would not exist and the error estimates would collapse.

Editorial extensions

If this is right

For defocusing smooth data, keeping only the leading profile already yields a rigorous O(ε²) approximation in H^{s-8} over times of order ε^{-2/(N+1)}, so reduced Schrödinger models inherit a concrete error bound.
Increasing regularity allows arbitrary-order approximations: including profiles up to order K lowers the error to ε^{K+1}, matching numerical observations of higher-order corrections.
The convergence holds in Sobolev spaces, not just L², so derivatives of the solution are also controlled by the Schrödinger approximation.
For focusing nonlinearity, the NLS description is validated up to the time when the Schrödinger solution itself may blow up, setting a fundamental limit on the non-relativistic approximation.
With less regular initial data the rate drops to O(ε), giving an explicit regularity-versus-rate tradeoff.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the quoted structural WKB identities hold, the same stability framework should extend to related two-dimensional dispersive limits, such as Klein-Gordon–Zakharov systems, with the decay rates of the limit system entering the prefactor.
The algebraic growth (1+t)^N likely reflects the borderline t^{-1} decay of the two-dimensional cubic NLS; longer time scales would require additional structure such as scattering or weighted estimates.
The arbitrary-order result suggests a quantitative numerical recipe: schemes that include the ε²u2 correction should see the error drop by two powers of ε, a directly testable prediction.
Because the NLS decay prefactor depends on the profile of the initial datum, two initial data with identical norms can produce very different error constants, so numerical comparisons should report the full profile-dependence.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The new content here is real: 2D cubic Klein-Gordon with H^s error estimates of order C(1+t)^N ε^2 over times of order ε^{-2/(N+1)}, plus an arbitrary-order WKB approximation. That is not in the earlier 3D paper [7] (linear prefactor) nor in Wu-Lei's 2D L^2 result [22]. The defocusing-case long-time bounds and the explicit polynomial-in-time prefactors are the genuine contribution. The proof strategy—reformulate as a symmetric hyperbolic system, build a WKB approximate solution, then run a stability argument against the cubic NLS decay—is sensible and carried out in reasonable detail for the stability part.

The main soft spot is exactly what the stress test flags: Propositions 3.4, 3.5, and Corollary 3.6, which give the algebraic form of the profiles U_{n,p} and the recursion relations, are stated without proof and deferred to [7]. These lemmas are load-bearing: the approximate solution U_a is constructed by induction through them, and the residual estimates and stability theorems presuppose that construction. The 2D system has 4x4 matrices, whereas [7] works with 5x5; the scalar profile equations are likely the same, but the paper does not demonstrate the transfer. If any structural identity fails in 2D, Theorem 2.2 has no object to control and Theorem 1.3 collapses. That is a genuine gap in self-containedness, not a manufactured one.

That said, the gap is plausibly fillable. The computations in Sections 3.1–3.5 explicitly work out orders up to O(ε^2) in 2D, matching the pattern of [7], and the deferred lemmas look like the same induction in a different ambient dimension. The paper also has some minor missing details, like the 'straightforward' corollary proofs in Section 4.4, but those are genuinely routine. No fatal internal inconsistency. I disagree with any claim that the self-citation is itself disqualifying—[7] is a published, peer-reviewed paper, and deferring to it is legitimate, provided the transfer to 2D is made explicit.

Who is this for? Researchers working on relativistic-to-nonrelativistic limits, especially numerical analysts who need rigorous error bounds matching observed (1+t)^N behavior. The paper deserves serious refereeing, but the referee should verify—or the authors should provide—the 2D versions of Propositions 3.4 and 3.5. I would recommend peer review with a request for revision: either prove the structural lemmas in an appendix or give a precise reduction to [7] showing the dimension change leaves the algebra unchanged.

Referee Report

1 major / 3 minor

Summary. The paper studies the non-relativistic limit ε→0 for the two-dimensional cubic Klein-Gordon equation (1.1)–(1.2). It rewrites the equation as the symmetric hyperbolic system (2.2) and constructs WKB approximate solutions of arbitrary even order. The main results are stability estimates for these approximate solutions in H^{s-2K-4} over long time intervals, yielding error O(ε^{K+1}) with polynomial-in-time prefactors. Theorems 1.3 and 1.4 give the convergence to the cubic nonlinear Schrödinger equation with rates O(ε^2) and O(ε), with time spans of size O(ε^{-2/(N+1)}) and O(ε^{-α/2}), respectively. The proof extends the three-dimensional analysis in [7] to two dimensions; the new ingredients are the defocusing long-time stability argument and the translation of the WKB construction to the 4×4 first-order system. The exposition is generally clear, but several load-bearing structural propositions are deferred to [7] without a self-contained verification for the 2D system.

Significance. If the results are correct, they improve on the earlier L^2 bound of Wu and Lei by providing H^s error estimates with explicit algebraic-in-time prefactors, and they are consistent with numerical observations. The stability framework and the careful tracking of initial-data-dependent growth exponents are valuable. The paper also makes a sharp distinction between high- and low-regularity data and between focusing and defocusing cases. However, the core WKB existence result is not established inside the paper: Propositions 3.4, 3.5 and Corollary 3.6 are stated without proofs and are deferred to [7], a paper that treats the three-dimensional case. Since all subsequent error estimates are conditional on the exact algebraic form of the approximate solution, the current manuscript is not self-contained on this point.

major comments (1)

[§3.6, equations (3.44)–(3.45)] The derivation of U_{4,3} and U_{4,5} is omitted with the note that the calculation is essentially the same as in [7]. These formulas feed into Corollary 3.6 and the regularity estimates, so they are part of the same gap. While it is reasonable to cite a prior paper for a routine computation, the dimension change makes it necessary to at least display the verification for the 2D case, or to state clearly that [7] already contains the 2D version. Without this, the induction is not independently checkable.

minor comments (3)

[§4.4] The deductions of Theorems 1.3–1.6 from Theorems 2.3–2.5 are said to be 'straightforward' and are omitted. Theorem 1.3 is shown explicitly, but a short paragraph for Theorems 1.4 and 1.6 would improve readability, particularly because the approximate solution u_a in Theorem 1.5 contains ε^K u_K and ε^{K+2}u_{K+2} and one must check that these extra terms are absorbed in the error.
[Title page and affiliation] There are typos in the affiliation line: 'school of Mathmatics' and 'Sistrict' should be 'School of Mathematics' and 'District'.
[§3.3, equation (3.22)] The last equality in (3.22) uses the Schrödinger equation for g0, but this is not stated at that point. Adding a parenthetical remark would help the reader.

Circularity Check

2 steps flagged · score 4.0 of 10

Arbitrary-order WKB existence is deferred to [7] by the same first author; the main error estimates inherit this load-bearing self-citation, but the stability/NLS-decay core is independent.

self citation load bearing [Section 3.6.2, Proposition 3.4]
"The proof of this proposition can be found in Section 3 of [7]. We omit the details for brevity."

Proposition 3.4 supplies the structural form of U_{n,±1} and the profile equations (3.47) for all n up to K+2. These equations are the induction backbone for the WKB approximate solution U_a used in Theorems 2.1/2.2. The paper explicitly does not prove this induction in the present 2D setting and instead refers to [7], a prior article co-authored by Y. Lu. The stability theorems 2.3–2.5 and the final error estimates 1.3–1.6 are all estimates of perturbations of this U_a; if the deferred identities fail, no residual bound (2.7) exists. Thus the load-bearing premise is justified only by an author-overlapping citation, without demonstrating that the 3D matrix identities carry over to the 4×4 2D system.
self citation load bearing [Section 3.6.4, Proposition 3.5 and Corollary 3.6]
"Proposition 3.5 can be proved by induction, as showed in [7]. ... By induction, we can deduce (see [7]) from Proposition 3.5 that:"

Proposition 3.5 and Corollary 3.6 give the recursion (3.48) and polynomial dependence of U_{n,p} on g_0,g_2,...; these determine the regularity and time-growth bounds used in Proposition 3.9 and in the residual estimate of Theorem 2.2. The proof is again deferred to [7], and the 2D case is asserted rather than established. Since Theorem 1.3/1.5/1.6 are corollaries of the stability of this U_a, the central derivation chain depends on this unverified self-citation, although the later stability analysis itself is not circular.

full rationale

The paper's main error estimates are not obtained by fitting a parameter and then calling it a prediction, nor by defining the target in terms of the input. The leading profile g0 is derived from the WKB equations and coincides with the cubic NLS solution with an independently cited well-posedness/decay theory (Bourgain; Hayashi–Tsutsumi). The stability estimates for U−U_a use genuine energy arguments and are not pre-baked. The circularity burden is concentrated in the construction of the arbitrary-order WKB approximate solution: the key structural Propositions 3.4, 3.5 and Corollary 3.6 are stated without proof and deferred to [7], a paper with overlapping authorship. These propositions are load-bearing because every later theorem estimates perturbations of the U_a whose existence depends on them, and the paper does not show the 2D 4×4 system satisfies the 3D identities. That is a self-citation load-bearing step, not a fully independent proof. However, this is not a case where the final prediction reduces by construction to the fit, so the score is moderate (4) rather than high.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central claim rests on known NLS well-posedness and decay (external), standard symmetric hyperbolic theory, and, crucially, on the WKB structural lemmas taken from the authors' prior work [7]. No new physical or mathematical entities are introduced. The only 'parameter' is the growth exponent N_φ,ψ, which is defined from existing constants rather than fitted.

free parameters (1)

N_φ,ψ (growth exponent in error bounds) = max{N̂, Ñ}, with N̂ from (3.69) and Ñ from (4.8)
Introduced to control the polynomial-in-time growth of the WKB remainder via NLS decay constants C1,C2 from Proposition 1.2. It sets the long-time scale T0 ε^{-2/(N+1)} in Theorems 1.3, 1.5, 2.4. Not fitted to data, but non-explicit and initial-data dependent.

assumptions (4)

domain assumption Local well-posedness of the 2D cubic NLS (Proposition 1.1)
Used to define g0 on [0,T*) for the focusing case and [0,∞) for the defocusing case.
domain assumption Global well-posedness and L∞ decay of the defocusing 2D cubic NLS (Proposition 1.2)
Provides the (1+t)^{-1} L∞ decay used throughout Sections 3.8 and 4.2 to control nonlinear interactions and derive the (1+t)^N prefactors.
domain assumption WKB expansion structure from [7] (Propositions 3.4, 3.5, Corollary 3.6)
The paper states these without proof, citing [7]; they are load-bearing for constructing the approximate solution u_a to arbitrary order.
standard math Symmetric hyperbolic well-posedness of (2.2)
Used in the stability arguments (Section 4) to obtain existence and uniqueness of the rescaled difference ˙U.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Error estimates in the non-relativistic limit for the two-dimensional cubic Klein-Gordon equation." pith.science (2026). https://pith.science/paper/GVZZF5TZ

@misc{pith2026250908271,
  author       = {Pith},
  title        = {Pith review of: Error estimates in the non-relativistic limit for the two-dimensional cubic Klein-Gordon equation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GVZZF5TZ}},
  note         = {Machine review of arXiv:2509.08271}
}

abstract

In this paper, we study the non-relativistic limit of the two-dimensional cubic nonlinear Klein-Gordon equation with a small parameter $0<\varepsilon \ll 1$ which is inversely proportional to the speed of light. We show the cubic nonlinear Klein-Gordon equation converges to the cubic nonlinear Schr\"{o}dinger equation with a convergence rate of order $O(\varepsilon^2)$. In particular, for the defocusing case with high regularity initial data, we show error estimates of the form $C(1+t)^N \varepsilon^2$ at time $t$ up to a long time of order $\varepsilon^{-\frac{2}{N+1}}$, while for initial data with limited regularity, we also show error estimates of the form $C(1+t)^M\varepsilon$ at time $t$ up to a long time of order $\varepsilon^{-\frac{1}{M+1}}$. Here $N$ and $M$ are constants depending on initial data. The idea of proof is to reformulate nonrelativistic limit problems to stability problems in geometric optics, then employ the techniques in geometric optics to construct approximate solutions up to an arbitrary order, and finally, together with the decay estimates of the cubic Schr\"{o}dinger equation, derive the error estimates.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages

[7]

W. Bao, Y. Lu, Z. Zhang, Convergence rates in the non-relativistic limit of the cubic Klein- Gordon equation, SIAM J. Math. Anal. 56 (2024), 6822–6860

work page 2024
[22]

Z. Lei, Y. Wu. Non-relativistic limit for the cubic nonlinear Klein-Gordon equation. arXiv: 2309.10235 (2023). 29

work page Pith review arXiv 2023
[1]

W. Bao, Y. Cai, X. Zhao. A uniformly accurate multiscale time integrator pseudospectral method for the Klein-Gordon equation in the non-relativistic limit regime. SIAM J. Numer. Anal. 52 (2014), 2488–2511

work page 2014
[2]

W. Bao, X. Dong. Analysis and comparison of numerical methods for the Klein-Gordon equation in the non-relativistic limit regime. Numer. Math. 120 (2012), 189–229

work page 2012
[3]

W. Bao, X. Zhao. A uniformly accurate multiscale time integrator spectral method for the Klein-Gordon-Zakharov system in the high-plasma-frequency limit regime. J. Comput. Phys. 327 (2016), 270–293

work page 2016
[4]

W. Bao, X. Zhao. A uniformly accurate (UA) multiscale time integrator Fourier pseudospectral method for the Klein-Gordon-Schr¨ odinger equations in the non-relativistic limit regime. Numer. Math. 135 (2017), 833–873

work page 2017
[5]

W. Bao, X. Zhao. Comparison of numerical methods for the nonlinear Klein-Gordon equation in the non-relativistic limit regime. J. Comput. Phys. 398 (2019). 28

work page 2019
[6]

W. Bao, X. Zhao. A uniformly second-order in time multiscale time integrator for the nonlinear Klein-Gordon equation in the non-relativistic limit regime. Preprint, 2019

work page 2019

Show all 22 references

[8]

Bourgain

J. Bourgain. Global solutions of nonlinear Schr¨ odinger equations. American Mathematical Society Colloquium Publications, 46, Amer. Math. Soc., Providence, RI, 1999

1999
[9]

C. Fan, Z. Zhao. Decay estimates for nonlinear Schr¨ odinger equations. Discrete Contin. Dyn. Syst. 41 (2021), 3973–3984

2021
[10]

Ginibre, G

J. Ginibre, G. Velo. The global Cauchy problem for the nonlinear Klein-Gordon equation. Math. Z. 189 (1985), 487–505

1985
[11]

Ginibre, G

J. Ginibre, G. Velo. The global Cauchy problem for the nonlinear Klein-Gordon equation. II, Ann. Inst. H. Poincar´ e Anal. Non Lin´ eaire 6 (1989), 15–35

1989
[12]

Hayashi, M

N. Hayashi, M. Tsutsumi.L ∞(Rn)-decay of classical solutions for nonlinear Schr¨ odinger equations. Proc. Roy. Soc. Edinburgh Sect. A 104 (1986), 309–327

1986
[13]

Y. Lu, Z. Zhang. Partially strong transparency conditions and a singular localization method in geometric optics, Arch. Ration. Mech. Anal. 222 (2016), 245–283

2016
[14]

Y. Lu, Z. Zhang. Higher order asymptotic analysis of the Klein-Gordon equation in the non- relativistic limit regime. Asymptot. Anal. 102 (2017), 157–175

2017
[15]

Masmoudi, K

N. Masmoudi, K. Nakanish. From nonlinear Klein-Gordon equation to a system of coupled nonlinear Schr¨ odinger equations. Math. Ann. 324 (2002), 359–389

2002
[16]

C. S. Morawetz, W. A. Strauss. Decay and scattering of solutions of a nonlinear relativistic wave equation. Comm. Pure Appl. Math. 25 (1972), 1–31

1972
[17]

Machihara

S. Machihara. The non-relativistic limit of the nonlinear Klein-Gordon equation. Funkcial. Ekvac. 44 (2001), 243–252

2001
[18]

Machihara, K

S. Machihara, K. Nakanishi, T. Ozawa. Nonrelativistic limit in the energy space for nonlinear Klein-Gordon equations, Math. Ann. 322 (2002), 603–621

2002
[19]

B. Najman. The nonrelativistic limit of the nonlinear Klein-Gordon equation, Nonlinear Anal. 15 (1990), 217–228

1990
[20]

W. A. Strauss, L. V´ azquez Mart ´ ınez. Numerical solution of a nonlinear Klein-Gordon equation, J. Comput. Phys. 28 (1978), 271–278

1978
[21]

Schratz, X

K. Schratz, X. Zhao. On comparison of asymptotic expansion techniques for nonlinear Klein- Gordon equation in the non-relativistic limit regime. Discrete Contin. Dyn. Syst. Ser. B 25 (2020), 2841–2865

2020

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [7]

W. Bao, Y. Lu, Z. Zhang, Convergence rates in the non-relativistic limit of the cubic Klein- Gordon equation, SIAM J. Math. Anal. 56 (2024), 6822–6860

work page 2024

[2] [22]

Z. Lei, Y. Wu. Non-relativistic limit for the cubic nonlinear Klein-Gordon equation. arXiv: 2309.10235 (2023). 29

work page Pith review arXiv 2023

[3] [1]

W. Bao, Y. Cai, X. Zhao. A uniformly accurate multiscale time integrator pseudospectral method for the Klein-Gordon equation in the non-relativistic limit regime. SIAM J. Numer. Anal. 52 (2014), 2488–2511

work page 2014

[4] [2]

W. Bao, X. Dong. Analysis and comparison of numerical methods for the Klein-Gordon equation in the non-relativistic limit regime. Numer. Math. 120 (2012), 189–229

work page 2012

[5] [3]

W. Bao, X. Zhao. A uniformly accurate multiscale time integrator spectral method for the Klein-Gordon-Zakharov system in the high-plasma-frequency limit regime. J. Comput. Phys. 327 (2016), 270–293

work page 2016

[6] [4]

W. Bao, X. Zhao. A uniformly accurate (UA) multiscale time integrator Fourier pseudospectral method for the Klein-Gordon-Schr¨ odinger equations in the non-relativistic limit regime. Numer. Math. 135 (2017), 833–873

work page 2017

[7] [5]

W. Bao, X. Zhao. Comparison of numerical methods for the nonlinear Klein-Gordon equation in the non-relativistic limit regime. J. Comput. Phys. 398 (2019). 28

work page 2019

[8] [6]

W. Bao, X. Zhao. A uniformly second-order in time multiscale time integrator for the nonlinear Klein-Gordon equation in the non-relativistic limit regime. Preprint, 2019

work page 2019