Stability of Stochastically Forced Solitons in the Korteweg-de Vries Equation

Hermen Jan Hupkes; Rik W.S. Westdorp

REVIEW 2 major objections 4 minor 29 references

Stability of Stochastically Forced Solitons in the Korteweg-de Vries Equation

T0 review · 2 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read This paper proves that stochastically forced KdV solitons remain close to the traveling-wave family on timescales where the deterministic forcing changes the amplitude by an O(1) factor.

desk verdict Real advance in stochastic KdV soliton stability, but S1 needs more regularity and Theorem 1.2's bound drops a T factor. read the letter →

arxiv 2504.17407 v1 pith:ZM443GWS submitted 2025-04-24 math.AP math.PR

Rik W.S. Westdorp , Hermen Jan Hupkes This is my paper · ORCID

classification math.APmath.PR MSC 60H1535Q5335C08

keywords stochastictravelingwavesKorteweg-deVriesequationstabilityforcingmultiplicativenoisesolitonmodulationweightedSobolevspacestranslation-invariant

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

Stochastically and deterministically forced Korteweg-de Vries solitons are shown to stay close to the family of traveling waves, even while the effective soliton amplitude drifts by a nontrivial amount. The central bound is an exit-time estimate: with small noise σ and small forcing ε, the probability that the perturbation leaves a weighted H¹ neighborhood before time T is at most CTσ²log(1/σ)+$CTe^{{-δη²/σ²}}$. Earlier stochastic stability results applied only on times small compared to $σ^{{-2}}$, where amplitude motion was negligible; here the deterministic forcing is allowed to change the amplitude by an O(1) factor within the stability window. The paper also proves that the amplitude is accurately described, up to small-probability error, by an explicit effective stochastic differential equation.

What carries the argument

The argument is carried by a variational-phase modulation decomposition together with a frozen-linearization step. The solution is written as a modulated soliton plus a remainder v, with orthogonality conditions ⟨v,φ_{c(t)}⟩=⟨v,ζ_{c(t)}⟩=0, where ζ_c=∫_{-∞}^x ∂_c φ_c dy; the modulation parameters c(t) and ξ(t) obey stochastic differential equations. On each short interval [T,T+ΔT] the paper freezes the amplitude at c(T) and studies a local remainder v_T against the semigroup generated by the linearized KdV operator L_{c(T)}. In exponentially weighted L² spaces this semigroup has a spectral gap of size w(c-w²) after removing two neutral modes, giving the exponential decay estimates (2.3)-(2.4) that damp the local remainder. Gaussian tail estimates for stochastic convolutions control the noise integrals on each interval, and a bootstrap over the partition proves the global exit-time bound.

What would settle it

Take f≡1, choose any even q∈H¹∩L¹ with nonnegative Fourier transform, integrate (1.1) numerically from φ_{c*} and measure P[t_st(η)<T] at T=$σ^{{-1}}$ for small σ; the theorem predicts this probability is at most Cσ log(1/σ)+$Ce^{{-δη²/σ²}}$, so observing a probability bounded away from zero as σ→0 at fixed η would falsify Theorem 1.1. A cheaper test targets Proposition 7.1: with the same setup, measure the first time the amplitude exits [cmin,cmax]; the bound says this exit probability before T=$σ^{{-1}}$ is at most $e^{{-δ17/σ}}$, which should be tiny.

Watch

Extended reading notes

Core claim

Under assumptions S1-S2, for any initial soliton φ_{c*} and any admissible deterministic forcing with ε∫|f|≤E, the modulation decomposition u(t,x+ξ(t))=φ_{c(t)}(x)+v(t,x) satisfies the exit-time bound P[t_st(η)<T]≤CTσ²log(1/σ)+$CTe^{{-δη²/σ²}}$, where t_st(η) is the first time the weighted H¹_w norm of v exceeds η. The two terms have distinct origins: the exponential term comes from linear damping on exponentially weighted spaces, while the σ²log(1/σ) term comes from the Itô drift growth of the unweighted L² norm combined with amplitude fluctuations. A companion theorem shows the reduced amplitude SDE (1.7) tracks the true amplitude for times up to T with error probability at most Cσ²/λ log(1/σ). The result establishes orbital stability of KdV solitons under combined deterministic and stochastic multiplicative forcing on timescales where the forced amplitude change is O(1).

Load-bearing premise

The proof needs the soliton amplitude c(t) to remain inside a fixed band [cmin,cmax] with w<√cmin/3, so that the spectral gap b<w(c-w²) stays strictly positive over the whole time interval; if the noise or forcing pushes c(t) out of this band, the linear decay estimates lose their constants and the iterative stability argument collapses.

Editorial extensions

If this is right

The bound P[t_st(η)<T]≤CTσ²log(1/σ)+CTe^{-δη²/σ²} makes stable soliton motion a high-probability event for T small compared with σ^{-2}log(1/σ)^{-1}.
The reduced amplitude SDE is quantitatively reliable: the probability that |c(t)-c_ap(t)| exceeds λ before time T is controlled by (Cσ²/λ)log(1/σ), so one can predict amplitude drift without resolving the dispersive radiation.
The deterministic forcing is allowed to change the amplitude by an O(1) factor during the stability window, meaning the result covers soliton amplification and decay rather than only infinitesimal fluctuations.
Stability can be propagated over many short intervals of length ΔT without needing a uniform-in-time linearization of the time-varying soliton, because each interval uses the frozen semigroup at its left endpoint.

Reading between the lines

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

An implication the authors leave implicit is that the log(1/σ) penalty comes from unweighted energy control; replacing that step by direct pointwise dispersive estimates should extend the result to times of order σ^{-2} without changing the rest of the architecture.
The frozen-linearization scheme should transfer to other solitary-wave PDEs whose linearized operators have a spectral gap in weighted spaces, since the components specific to KdV are the explicit soliton family and the L¹_w nonlinearity estimate.
Theorem 1.2 makes the reduced amplitude SDE a predictive tool: given the noise kernel q, one can compute g_Q and estimate when amplitude fluctuations become observable, before dispersive radiation has grown.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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

This is a genuine step forward: the first rigorous stability theorem for KdV solitons under stochastic forcing that allows the amplitude to change by an O(1) factor over the time horizon. The piecewise local modulation setup is clever, the deterministic/stochastic decomposition is worked out in detail, and the proof of Theorem 1.1 through Section 8 is internally coherent. The citations to [27] and [28] are to companion works, not to the target result, and there is no circularity in the argument.

Two problems stand out, and one of them is in the main theorem. First, S1's assumption q ∈ H^1 ∩ L^1 is not enough for Lemma 5.4. The Hilbert-Schmidt identity in that lemma contains a term involving the derivative of q_{1/2}; it is finite only if the integral of ω^2 times the Fourier transform of q is finite. The stress-test example—hat q(ω) = c(1+ω^2)^{-1}(log(e+ω^2))^{-3/4}—satisfies S1 but has infinite second moment, so the operator hT_ξ is not Hilbert-Schmidt into H^1_w in general. The inequality in Lemma 5.4, and hence Lemma 5.5 and Proposition 5.1, fail as stated. The fix is straightforward: strengthen S1 to require that integral to be finite, i.e. q_{1/2} ∈ H^1. This is a missing hypothesis in the statement of the main theorem, not a cosmetic tightening.

Second, Theorem 1.2 is not supported. Lemma 9.3 carries an unstated restriction σ√T ≤ const from the Gronwall step, and its bound contains an e^{σ^2T} factor. In the proof of Theorem 1.2, the T factor is then dropped when arriving at (1.10). For fixed λ > 0 and large T, the exit probability for tap should tend to 1, so a T-independent bound cannot hold. This looks fixable only by restating the theorem with a time restriction or a T-dependence; as written, it is false.

The core idea is good and the main stability result will likely survive once the regularity assumption is added. Both gaps, however, need to be addressed before the statements can be trusted. I agree with the reader's conditional verdict, and the stress-test concern is real.

This paper is for people working on stochastic dispersive PDEs and phase-tracking methods. It deserves a serious referee, but the refereeing process should demand a corrected S1 and a repaired Theorem 1.2.

Referee Report

2 major / 4 minor

Summary. The manuscript studies the stochastic Korteweg-de Vries equation with multiplicative, spatially homogeneous noise and deterministic forcing. It introduces a modulation decomposition around the soliton family, derives an effective SDE for the soliton amplitude, and proves two main results: Theorem 1.1, a weighted H^1 exit-time bound showing that the solution remains close to the modulated soliton with high probability, and Theorem 1.2, a bound on the time during which the actual amplitude is approximated by the reduced SDE. The proof strategy partitions time into intervals of length Delta_T, uses Pego-Weinstein semigroup estimates in exponentially weighted spaces, controls the unweighted L^2 growth by energy arguments, and combines these with Gaussian tail estimates for stochastic convolutions.

Significance. If the results are correct as stated, the paper is a significant extension of previous soliton-stability results for stochastic KdV, since it allows the deterministic forcing to produce O(1) amplitude modulation rather than only small fluctuations. The proof is substantial and does not rely on fitting or tuning of constants: the modulation system is derived explicitly and the stability argument is carried out through a bootstrap over local intervals. The main technical tool, namely freezing the linearized operator on short intervals and using the resulting spectral gap, is natural and is presented in considerable detail. However, two load-bearing issues need to be resolved before the stated theorems can be accepted.

major comments (2)

[§2 (S1); §5, Lemma 5.4] The standing assumption S1 only requires q ∈ H^1 ∩ L^1 with ŵhat q ≥ 0. These conditions do not imply q_{1/2} ∈ H^1. For example, ŵhat q(ω) = c(1+ω^2)^{-1}(log(e+ω^2))^{-3/4} satisfies q ∈ H^1 ∩ L^1 and ŵhat q ≥ 0, but ∫ ω^2 ŵhat q(ω)dω = ∞. The Hilbert-Schmidt identity displayed in Lemma 5.4 contains the term ‖q'_{1/2}‖_{L^2}^2 ‖h‖_{L^2_w}^2, which is infinite for such q and generic h ∈ H^1_w. Since Lemma 5.4 is used in Lemma 5.5 and then in Proposition 5.1, the proof of Theorem 1.1 is not justified under S1 as stated. A strengthened hypothesis such as ∫(1+ω^2)ŵhat q(ω)dω < ∞ (equivalently q_{1/2} ∈ H^1) appears necessary; the authors should add it to S1 and verify that the cited well-posedness results hold under this strengthened assumption.
[§9, Theorem 1.2 and Lemma 9.3] The conclusion (1.10) is not supported by the proof. Lemma 9.3 produces, after imposing the restriction σ√T ≤ 1/(2C_{21}) inside its proof, a bound of the form C_{22} T e^{σ^2 T} η^2/λ. In the proof of Theorem 1.2 this is combined with Theorem 1.1, yielding in particular terms with an explicit factor T; the final step then drops every T factor and also drops the e^{σ^2 T} factor when claiming (1.10). For T ≥ 1 this is not a legitimate manipulation. Moreover, the left-hand side of (1.10) is nondecreasing in T while the right-hand side is independent of T, so the asserted bound cannot hold for large T unless the approximation error never exits the interval; the paper provides no argument for such a conclusion. Theorem 1.2 should be restated with the T-dependent bound actually proved, together with the hypotheses on σ√T, or the claim should be removed.

minor comments (4)

[§2, S1] In the displayed definition of Q, the integral is written as ∫ q(x−y) f(y) dx; the integration variable should be dy.
[§8, Proof of Theorem 1.1] The proof states 'For ease of exposition, we consider T ≥ 1 for which T/ΔT ∈ N'. The reduction to such T is not explained; either the statement should mention that T can be enlarged to the next multiple of ΔT, or the final partial interval should be handled explicitly.
[Appendix B, Proof of Lemma 3.2] The mild Itô formula computation is very dense. A short remark identifying the exact version of the mild Itô formula used (reference [6, Theorem 1]) and explaining how the formal strong-form computation is justified would improve readability.
[Abstract and title] There is a typo in the abstract: 'Kortew eg-de Vries' should read 'Korteweg-de Vries'. Several occurrences of the notation c_{ap} and Ω_{ap} are typeset inconsistently (appearing as 'cap' and 'Omega ap').

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: modulation equations are derived, not fitted, and the stability proof relies on external semigroup and well-posedness results.

full rationale

The claimed derivation chain is self-contained rather than circular. The reduced amplitude process c_ap in (1.7) is obtained by evaluating the rigorously derived modulation system at v=0 in Section 3, with gQ(c)=cd(0,c) computed from the modulation coefficients; it is an output of the derivation, not an input fitted to the target result. Theorem 1.2 then proves the validity of this approximation via the SDE (9.1) and a Gronwall argument, rather than assuming it. The main stability estimate, Theorem 1.1, is built from the Pego-Weinstein semigroup bounds (Theorem 2.2, cited to [25,23]), the local modulation estimates of Sections 4-6, and the Gaussian tail bound Theorem 5.3; no target probability bound is inserted as an assumption. Self-citations to [27] and [28] are used for notation, the S1 stochastic set-up, and deterministic well-posedness in Lemma 2.1; these are prior published results that do not contain Theorem 1.1, so they are not load-bearing in a circular way. The paper explicitly acknowledges the time-scale limitation "At present, we can hence not carry out our arguments on a time-interval [0,Tmax sigma^{-2}]" in Section 1, which is an honest scope restriction rather than a disguised circular step. A possible regularity gap in Lemma 5.4 concerning whether q_{1/2} belongs to H^1 under S1 would be a correctness issue, not a constructional circularity, since it does not arise from defining the target through the hypothesis. Overall, no step in the derivation reduces by definition to its own inputs.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No parameters are fitted to data; all constants are universal analytic choices. The central claim depends on the stated modeling assumptions and standard analytic tools listed above.

assumptions (5)

domain assumption S1: q in H^1(R) intersect L^1(R), q even and nonnegative Fourier transform; f bounded with |f| <= 1; the SPDE (1.1) has a unique mild solution in L^2(Omega; C(R+; L2)).
Sets up the noise covariance and well-posedness; taken from [27, Section 2.1.2] and [8, 28].
standard math Pego-Weinstein semigroup bounds (2.3)-(2.4): for b < w(c-w^2), the projected linearized flow e^{L_c t} Q_c decays exponentially on L^2_w and L^1_w with explicit singular rates.
Central linear stability input; quoted from [25] and [23].
standard math Gaussian tail inequality for stochastic convolutions (Theorem 5.3), proved in Appendix B via maximal inequalities from [24].
Used to bound the probability that weighted norms of stochastic convolutions exceed a threshold.
standard math The mild Ito formula of Da Prato, Jentzen and Roeckner [6] applies to the modulation functional F(u,c,xi).
Justifies the formal modulation SDEs; the application is technical and only sketched in Appendix B.
domain assumption Condition C2: cmin <= c* exp(-3E), cmax >= c* exp(3E), f in L^1(0,infty), and epsilon integral |f| <= E.
Keeps the soliton amplitude inside the band where the spectral gap is uniform; needed for Proposition 7.1.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Stability of Stochastically Forced Solitons in the Korteweg-de Vries Equation." pith.science (2026). https://pith.science/paper/ZM443GWS

@misc{pith2026250417407,
  author       = {Pith},
  title        = {Pith review of: Stability of Stochastically Forced Solitons in the Korteweg-de Vries Equation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZM443GWS}},
  note         = {Machine review of arXiv:2504.17407}
}

read the original abstract

We study the stability and dynamics of solitons in the Korteweg-de Vries (KdV) equation in the presence of noise and deterministic forcing. The noise is space-dependent and statistically translation-invariant. We show that, for small forcing, solitons remain close to the family of traveling waves in a weighted Sobolev norm, with high probability. We study the effective dynamics of the soliton amplitude and position via their variational phase, for which we derive explicit modulation equations. The stability result holds on a time scale where the deterministic forcing induces significant amplitude modulation.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 27 canonical work pages

[27]

Westdorp and H.J

R.W.S. Westdorp and H.J. Hupkes. Long-timescale soliton dynamic s in the Korteweg–de Vries equation with multiplicative translation-invariant noise. Physica D. , 460:134065, 2024

work page 2024
[28]

Westdorp and H.J

R.W.S. Westdorp and H.J. Hupkes. Soliton Ampliﬁcation in the Korte weg-de Vries Equation by Multiplicative Forcing. Commun. Pure Appl. Anal. , 2025

work page 2025
[1]

Adams and J

Z.P. Adams and J. MacLaurin. The isochronal phase of stochast ic pde and integral equations: Metastability and other properties. J. Diﬀer. Equ. , 414:773–816, 2025

work page 2025
[2]

van den Bosch and H.J

M. van den Bosch and H.J. Hupkes. Multidimensional Stability of Plan ar Travelling Waves for Stochastically Perturbed Reaction-Diﬀusion Systems. arXiv preprint arXiv:2406.04232 , 2024

arXiv 2024
[3]

Local Phase Tracking and Metastability of Planar Waves in Stochastic Reaction-Diffusion Systems

M. van den Bosch and H.J. Hupkes. Local Phase Tracking and Met astability of Planar Waves in Stochastic Reaction-Diﬀusion Systems. arXiv preprint arXiv:2504.09350 , 2025

work page Pith review arXiv 2025
[4]

Boussinesq

J. Boussinesq. Th´ eorie de l’intumescence liquide appel´ ee onde s olitaire ou de translation se propageant dans un canal rectangulaire. CR Acad. Sci. Paris , 72:755–759, 1871

work page
[5]

Cartwright and G.A

M. Cartwright and G.A. Gottwald. Collective coordinate framewor k to study solitary waves in stochastically perturbed Korteweg–de Vries equations. Phys. Rev. E , 104(2):024201, 2021

work page 2021
[6]

Da Prato, A

G. Da Prato, A. Jentzen, and M. R¨ ockner. A mild Itˆ o formula fo r SPDEs. Trans. Am. Math. Soc., 372(6):3755–3807, 2019

work page 2019

Show all 29 references

[7]

Da Prato and J

G. Da Prato and J. Zabczyk. Stochastic equations in inﬁnite dimensions . Cambridge university press, 2014

2014
[8]

De Bouard and A

A. De Bouard and A. Debussche. The Korteweg-de Vries equatio n with multiplicative homo- geneous noise. In Stochastic Diﬀerential Equations: Theory And Application s: A Volume in Honor of Professor Boris L Rozovskii , pages 113–133. World Scientiﬁc, 2007

2007
[9]

De Bouard and A

A. De Bouard and A. Debussche. Random modulation of solitons fo r the stochastic Korteweg-de Vries equation. In Annales de l’IHP Analyse non lin´ eaire, volume 24, pages 251–278, 2007

2007
[10]

De Bouard and A

A. De Bouard and A. Debussche. Soliton dynamics for the Korte weg-de Vries equation with multiplicative homogeneous noise. Electron. J. Probab., 14:1727–1744, 2009

2009
[11]

De Bouard and R

A. De Bouard and R. Fukuizumi. Modulation analysis for a stochas tic NLS equation arising in Bose–Einstein condensation. Asymptot. Anal. , 63(4):189–235, 2009

2009
[12]

Eichinger, M.V

K. Eichinger, M.V. Gnann, and C. Kuehn. Multiscale analysis for tr aveling-pulse solutions to the stochastic FitzHugh–Nagumo equations. Ann. Appl. Probab. , 32(5):3229–3282, 2022

2022
[13]

Gardner, J.M

C.S. Gardner, J.M. Greene, M.D. Kruskal, and R.M. Miura. Method for solving the Korteweg- deVries equation. Phys. Rev. Lett. , 19(19):1095, 1967. 39

1967
[14]

Gnann, R.W.S

M.V. Gnann, R.W.S. Westdorp, and J. van Winden. Solitary waves in a stochastic parametri- cally forced nonlinear Schr¨ odinger equation. arXiv preprint arXiv:2403.04625 , 2024

2024
[15]

Hamster and H.J

C.H.S. Hamster and H.J. Hupkes. Travelling waves for reaction–d iﬀusion equations forced by translation invariant noise. Physica D. , 401:132233, 2020

2020
[16]

R.L. Herman. The stochastic, damped KdV equation. J. Phys. A. Math. Gen. , 23(7):1063, 1990

1990
[17]

Korteweg and G

D.J. Korteweg and G. de Vries. On the change of form of long wav es advancing in a rectangular channel, and a new type of long stationary wave. Philos. Mag. , 39:422–443, 1895
[18]

Kruger and W

J. Kruger and W. Stannat. Front propagation in stochastic ne ural ﬁelds: a rigorous mathemat- ical framework. SIAM J. Appl. Dyn. Syst. , 13(3):1293–1310, 2014

2014
[19]

Liu and M

W. Liu and M. R¨ ockner. Stochastic partial diﬀerential equations: an introductio n. Springer, 2015

2015
[20]

MacLaurin

J. MacLaurin. Phase reduction of waves, patterns, and oscilla tions subject to spatially extended noise. SIAM J. Appl. Math. , 83(3):1215–1244, 2023

2023
[21]

Merle and L

F. Merle and L. Vega. L2 stability of solitons for KdV equation. Int. Math. Res. Not. , 2003(13):735–753, 2003

2003
[22]

Miura, C.S

R.M. Miura, C.S. Gardner, and M.D. Kruskal. Korteweg-de vries e quation and generalizations. II. Existence of conservation laws and constants of motion. J. Math. Phys. , 9(8):1204–1209, 1968

1968
[23]

Mizumachi and N

T. Mizumachi and N. Tzvetkov. L2-stability of solitary waves for the KdV equation via Pego and Weinstein’s method. RIMS Kˆ okyˆ uroku, B49:33–63, 2014

2014
[24]

van Neerven and M.C

J.M.A.M. van Neerven and M.C. Veraar. Maximal inequalities for sto chastic convolutions in 2-smooth Banach spaces and applications to stochastic evolution e quations. Philos. Trans. R. Soc. A , 378(2185):20190622, 2020

2020
[25]

Pego and M.I

R.L. Pego and M.I. Weinstein. Asymptotic stability of solitary wave s. Commun. Math. Phys. , 164(2):305–349, 1994

1994
[26]

Veraar and L

M.C. Veraar and L. Weis. A note on maximal estimates for stocha stic convolutions. Czechoslov. Math. J. , 61(3):743–758, 2011

2011
[29]

van Winden

J. van Winden. Noncommutative orbital stability of stochastic p atterns in Banach spaces. arXiv preprint arXiv:2406.16642, 2024. 40

2024 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [27]

Westdorp and H.J

R.W.S. Westdorp and H.J. Hupkes. Long-timescale soliton dynamic s in the Korteweg–de Vries equation with multiplicative translation-invariant noise. Physica D. , 460:134065, 2024

work page 2024

[2] [28]

Westdorp and H.J

R.W.S. Westdorp and H.J. Hupkes. Soliton Ampliﬁcation in the Korte weg-de Vries Equation by Multiplicative Forcing. Commun. Pure Appl. Anal. , 2025

work page 2025

[3] [1]

Adams and J

Z.P. Adams and J. MacLaurin. The isochronal phase of stochast ic pde and integral equations: Metastability and other properties. J. Diﬀer. Equ. , 414:773–816, 2025

work page 2025

[4] [2]

van den Bosch and H.J

M. van den Bosch and H.J. Hupkes. Multidimensional Stability of Plan ar Travelling Waves for Stochastically Perturbed Reaction-Diﬀusion Systems. arXiv preprint arXiv:2406.04232 , 2024

arXiv 2024

[5] [3]

Local Phase Tracking and Metastability of Planar Waves in Stochastic Reaction-Diffusion Systems

M. van den Bosch and H.J. Hupkes. Local Phase Tracking and Met astability of Planar Waves in Stochastic Reaction-Diﬀusion Systems. arXiv preprint arXiv:2504.09350 , 2025

work page Pith review arXiv 2025

[6] [4]

Boussinesq

J. Boussinesq. Th´ eorie de l’intumescence liquide appel´ ee onde s olitaire ou de translation se propageant dans un canal rectangulaire. CR Acad. Sci. Paris , 72:755–759, 1871

work page

[7] [5]

Cartwright and G.A

M. Cartwright and G.A. Gottwald. Collective coordinate framewor k to study solitary waves in stochastically perturbed Korteweg–de Vries equations. Phys. Rev. E , 104(2):024201, 2021

work page 2021

[8] [6]

Da Prato, A

G. Da Prato, A. Jentzen, and M. R¨ ockner. A mild Itˆ o formula fo r SPDEs. Trans. Am. Math. Soc., 372(6):3755–3807, 2019

work page 2019