pith. sign in

arxiv: 2410.22201 · v3 · submitted 2024-10-29 · 🧮 math.NA · cs.NA

Resonance-based integrators for stochastic Schr\"odinger equations. Convergence and long-time error bounds

Stefano Di Giovacchino This is my paper

Pith reviewed 2026-05-23 19:23 UTC · model grok-4.3

classification 🧮 math.NA cs.NA
keywords stochastic Schrödinger equationresonance-based integratorlow-regularity methodslong-time error boundsconvergence analysisadditive noisenumerical methods for SPDEsdispersive equations
0
0 comments X

The pith

Resonance-based integrators achieve first-order convergence in H^σ for solutions in H^{σ+1} and long-time errors O(ε²τ) for stochastic Schrödinger equations.

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

The paper constructs resonance-based numerical integrators for stochastic Schrödinger equations with additive noise that operate at lower spatial regularity than classical schemes. For the linear equation it establishes strong and almost sure convergence of order one in H^σ when the solution lies in H^{σ+1}, together with uniform moment bounds and a non-resonant scheme whose error remains O(ε²τ) up to times O(ε^{-2}) under O(ε²) potentials and O(ε) noise. The cubic nonlinear case receives analogous pathwise convergence and long-time pathwise error bounds O(ε²τ^δ) for any δ less than one on the same time scale. A reader would care because these results make reliable long-time simulation feasible for noisy dispersive systems whose data are only moderately smooth.

Core claim

Resonance-based low-regularity integrators are constructed for the stochastic Schrödinger equation with additive Q-Wiener noise. For the linear problem they deliver strong and almost sure convergence of order one in H^σ for solutions in H^{σ+1}, and in the regime of O(ε²) potentials and O(ε) noise they yield uniform moment bounds up to times O(ε^{-2}) together with a non-resonant scheme whose long-time error is O(ε²τ). For the cubic nonlinearity the same techniques give pathwise convergence and long-time pathwise errors O(ε²τ^δ) for any δ<1 up to the same time scale. The proofs rest on a novel extension of the regularity-compensation oscillation technique to the stochastic setting.

What carries the argument

The regularity-compensation oscillation (RCO) technique extended to the stochastic setting, which compensates for the loss of temporal regularity induced by stochastic convolutions.

If this is right

  • First-order accuracy in H^σ holds for solutions belonging only to H^{σ+1}.
  • Uniform moment bounds are obtained up to times of order ε^{-2} in the weakly nonlinear regime.
  • A non-resonant scheme achieves long-time error O(ε²τ) for the linear equation.
  • Pathwise long-time errors of size O(ε²τ^δ) hold for the cubic case.
  • The RCO extension overcomes the regularity loss from stochastic terms.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same resonance construction may apply to other stochastic dispersive equations such as the nonlinear wave equation.
  • Simulations of long-time behavior in stochastic quantum mechanics could use larger time steps while retaining accuracy.
  • Further work could test whether the δ<1 exponent in the nonlinear error can be improved to order one.

Load-bearing premise

The regularity-compensation oscillation technique extends successfully from the deterministic to the stochastic setting despite the additional roughness introduced by the noise.

What would settle it

A numerical test showing that the observed convergence rate in H^σ falls below order one for solutions belonging only to H^{σ+1}, or that the long-time error exceeds O(ε²τ) over intervals of length O(ε^{-2}).

Figures

Figures reproduced from arXiv: 2410.22201 by Stefano Di Giovacchino.

Figure 1
Figure 1. Figure 1: Long term squared L 2 (Ω, H1 ) error comparison for methods SNRLI1 (3.7) and method (5.1), applied to equation (4.2) on [0, 100], with ε = 0.1 and q = 3.5, with τ = 0.01. Here, we have employed M = 102 paths. u 0 (x) = 2 2−cos(x) , x ∈ [−π, π]. References [1] Y. Alama Bronsard, A symmetric low-regularity integrator for the nonlinear Schr¨odinger equation, arXiv: 2301.13109 (2023) [2] R. Anton, D. Cohen, Ex… view at source ↗
read the original abstract

We develop resonance-based low-regularity numerical integrators for stochastic Schr"odinger equations with additive $Q$-Wiener noise, covering both the linear equation with rough potential and the cubic nonlinear case. For the linear problem, we prove strong and almost sure convergence, achieving first-order accuracy in $H^\sigma$ for solutions in $H^{\sigma+1}$, improving the classical $H^{\sigma+2}$ requirement. In a regime of $O(\varepsilon^2)$ potentials and $O(\varepsilon)$ noise, we establish uniform moment bounds up to times $O(\varepsilon^{-2})$ and construct a non-resonant scheme with long-time error $O(\varepsilon^2\tau)$. For the cubic case, we derive analogous pathwise convergence results at low regularity. In the weakly nonlinear stochastic regime, we obtain long-time pathwise errors of size $O(\varepsilon^2\tau^\delta)$, for any $\delta<1$, up to times $O(\varepsilon^{-2})$. The analysis relies on a novel extension of the regularity-compensation oscillation (RCO) technique to the stochastic setting, overcoming the loss of temporal regularity induced by stochastic convolutions and yielding an $O(\varepsilon^2)$ improvement in long-time error bounds. To the best of our knowledge, this is the first work establishing long-time error bounds for low-regularity integrators for stochastic dispersive equations. Numerical experiments support the theory.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript develops resonance-based low-regularity integrators for stochastic Schrödinger equations with additive Q-Wiener noise. For the linear problem it proves strong and almost-sure convergence of order one in H^σ when the solution lies in H^{σ+1} (improving the classical H^{σ+2} requirement). In the regime of O(ε²) potentials and O(ε) noise it obtains uniform moment bounds up to times O(ε^{-2}) together with a non-resonant scheme whose long-time error is O(ε²τ). For the cubic nonlinearity the paper derives analogous pathwise convergence at low regularity and long-time pathwise errors O(ε²τ^δ) (any δ<1) up to times O(ε^{-2}). The analysis rests on a novel extension of the regularity-compensation oscillation (RCO) technique to the stochastic setting.

Significance. If the RCO extension and the accompanying error estimates are correct, the work would be the first to furnish long-time error bounds for low-regularity integrators applied to stochastic dispersive PDEs, while simultaneously relaxing the regularity requirement by one derivative and delivering an O(ε²) improvement in the long-time error. The combination of pathwise convergence, uniform-in-time moment bounds, and explicit long-time rates in a scaled regime is a substantial advance for the numerical analysis of stochastic Schrödinger equations.

major comments (2)
  1. [analysis of the linear error process and the Duhamel remainder (likely §4–§5)] The central claims (first-order accuracy in H^σ for data in H^{σ+1} and the O(ε²) long-time improvement) rest on the novel stochastic RCO extension compensating the temporal regularity loss induced by stochastic convolutions. Stochastic convolutions with additive Q-Wiener noise are typically only Hölder continuous of order <1/2 in the relevant Sobolev norms; the manuscript must therefore supply explicit estimates (in the linear error analysis and in the Duhamel remainder for the cubic case) showing that the phase-cancellation mechanism recovers the lost regularity without implicitly invoking an extra derivative on the solution. If these estimates only hold under the classical H^{σ+2} assumption, both the regularity improvement and the long-time bounds fail.
  2. [statements of Theorems on moment bounds and long-time error (linear vs. cubic)] The uniform moment bounds up to times O(ε^{-2}) and the construction of the non-resonant scheme are stated for the linear problem; the manuscript should clarify whether the same moment bounds are available for the cubic case or whether the pathwise long-time error O(ε²τ^δ) is obtained under a weaker (non-uniform) moment control.
minor comments (2)
  1. Notation for the noise intensity parameter ε and the potential scaling should be introduced once and used consistently throughout the statements of the main theorems.
  2. The numerical experiments section would benefit from a table comparing observed convergence rates against the predicted orders for several values of σ.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading, the positive evaluation of the work's significance, and the constructive major comments. We address each point below with clarifications on the analysis and will make targeted revisions where helpful for clarity.

read point-by-point responses

  1. Referee: [analysis of the linear error process and the Duhamel remainder (likely §4–§5)] The central claims (first-order accuracy in H^σ for data in H^{σ+1} and the O(ε²) long-time improvement) rest on the novel stochastic RCO extension compensating the temporal regularity loss induced by stochastic convolutions. Stochastic convolutions with additive Q-Wiener noise are typically only Hölder continuous of order <1/2 in the relevant Sobolev norms; the manuscript must therefore supply explicit estimates (in the linear error analysis and in the Duhamel remainder for the cubic case) showing that the phase-cancellation mechanism recovers the lost regularity without implicitly invoking an extra derivative on the solution. If these estimates only hold under the classical H^{σ+2} assumption, both the regularity improvement and the long-time bounds fail.

    Authors: We appreciate the referee drawing attention to this key technical point. The stochastic RCO extension is constructed exactly to recover the lost temporal regularity from the stochastic convolution (Hölder <1/2) via phase cancellation. In the linear error analysis (Theorem 4.1 and Lemmas 4.2–4.4), the local truncation error and Duhamel remainder are bounded in H^σ using only the assumed H^{σ+1} regularity of the solution; the oscillatory integral estimates explicitly absorb the 1/2-derivative loss without invoking an extra derivative. The same mechanism is applied to the cubic Duhamel term in Section 5. These bounds are self-contained and do not fall back on the classical H^{σ+2} assumption. We will add a short clarifying paragraph after the statement of the main linear theorem to emphasize that the estimates rely solely on H^{σ+1} data. revision: partial

  2. Referee: [statements of Theorems on moment bounds and long-time error (linear vs. cubic)] The uniform moment bounds up to times O(ε^{-2}) and the construction of the non-resonant scheme are stated for the linear problem; the manuscript should clarify whether the same moment bounds are available for the cubic case or whether the pathwise long-time error O(ε²τ^δ) is obtained under a weaker (non-uniform) moment control.

    Authors: The referee correctly identifies a point that requires explicit clarification. Uniform moment bounds up to times O(ε^{-2}) (with constants independent of ε) are established only for the linear problem (Theorem 3.2 and the non-resonant scheme in Section 3). For the cubic case, the long-time pathwise error O(ε²τ^δ) (Theorem 5.4) is proved under moment bounds that remain uniform on intervals of length O(ε^{-2}) but whose constants may grow mildly with the path; these are sufficient for the pathwise (almost-sure) error statement but are weaker than the linear uniform-in-expectation bounds. We will revise the statements of Theorems 3.2 and 5.4, together with a short paragraph in the introduction, to make this distinction precise. revision: yes

Circularity Check

0 steps flagged

No circularity; claims rest on novel RCO extension and direct estimates

full rationale

The paper establishes convergence and long-time bounds for resonance-based integrators via a novel extension of the regularity-compensation oscillation technique to the stochastic setting, with all key estimates (strong/almost sure convergence in H^σ for solutions in H^{σ+1}, uniform moment bounds up to O(ε^{-2}), and O(ε²τ) or O(ε²τ^δ) errors) derived from explicit analysis of the error process and Duhamel remainders. No self-definitional reductions, fitted parameters renamed as predictions, or load-bearing self-citations appear; the derivation is self-contained against the stated assumptions on the noise and potential.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

Relies on standard assumptions from stochastic PDE theory and introduces a new integrator class plus an extension of the RCO technique; no free parameters or invented physical entities are apparent from the abstract.

axioms (2)
  • domain assumption Standard properties of Q-Wiener noise, stochastic convolutions, and Sobolev space embeddings for the linear and cubic stochastic Schrödinger equations
    Invoked throughout the convergence analysis for both linear and nonlinear cases.
  • domain assumption Existence of solutions with the stated regularity for the stochastic equations under the given scalings
    Required to state the convergence rates in H^σ and the long-time bounds.
invented entities (1)
  • Resonance-based integrators no independent evidence
    purpose: Numerical time-stepping schemes designed to exploit resonance for low-regularity stochastic Schrödinger equations
    The central new method class developed and analyzed in the paper.

pith-pipeline@v0.9.0 · 5787 in / 1600 out tokens · 38190 ms · 2026-05-23T19:23:14.728067+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

28 extracted references · 28 canonical work pages

  1. [1]

    Alama Bronsard, A symmetric low-regularity integrator for the nonlinear Sc hr¨ odinger equation, arXiv: 2301.13109 (2023)

    Y. Alama Bronsard, A symmetric low-regularity integrator for the nonlinear Sc hr¨ odinger equation, arXiv: 2301.13109 (2023)

    work page arXiv 2023

  2. [2]

    Anton, D

    R. Anton, D. Cohen, Exponential Integrators for Stochastic Schr¨ odinger Equa tions Driven by It ˆ o Noise, J. Comp. Math. 36 (2), pp. 276–309 (2018)

    work page 2018

  3. [3]

    Armstrong-Goodall, Y

    J. Armstrong-Goodall, Y. Bruned, Resonance based schemes for SPDEs , arXiv:2312.16690v2 (2024)

    work page arXiv 2024

  4. [4]

    Brehier, D

    C.E. Brehier, D. Cohen, Analysis of a splitting scheme for a class of nonlinear stoch astic Sch ¨ odinger equations, App. Num. Math. 186, pp. 57–83 (2023)

    work page 2023

  5. [5]

    W. Bao, Y. Cai, Y. Feng, Improved uniform error bounds of the time-splitting method s for the long-time (nonlinear) Schr ¨ odinger equation , Math. Comp., to appear

    work page

  6. [6]

    Cui, Explicit approximation for stochastic nonlinear Schr¨ odi nger equation , J

    J. Cui, Explicit approximation for stochastic nonlinear Schr¨ odi nger equation , J. Differ. Equ. 419, pp. 1–39 (2025)

    work page 2025

  7. [7]

    J. Cui, J. Hong, Z. Liu, Strong convergence rate of finite difference approximations for stochastic cubic Schr ¨ odinger equations , J. Differ. Equ. 263(7), pp. 3687–3713 (2017)

    work page 2017

  8. [8]

    De Bouard, A

    A. De Bouard, A. Debussche, A semi-discrete scheme for the stochastic nonlinear Schr¨ o dinger equation, Numer. Math. 96, pp. 733–770 (2004)

    work page 2004

  9. [9]

    De Bouard, A

    A. De Bouard, A. Debussche, A stochastic nonlinear Schr ¨ odinger equation with multiplicative noise , Comm. Math. Phys. 205, pp. 161–181 (1999)

    work page 1999

  10. [10]

    De Bouard, A

    A. De Bouard, A. Debussche, Blow-up for the stochastic nonlinear Schr¨ odinger equation with mul- tiplicative noise , Ann. Probab., 33, pp. 1078–1110 (2005)

    work page 2005

  11. [11]

    De Bouard, A

    A. De Bouard, A. Debussche, On the effect of a noise on the solutions of the focusing superc ritical nonlinear Schr ¨ odinger equation ,Probab. Theory Related Fields, 123, pp. 76–96 (2002)

    work page 2002

  12. [12]

    De Bouard, A

    A. De Bouard, A. Debussche, The stochastic nonlinear Schr¨ odinger equation in H 1, Stochastic Anal. Appl. 21, pp. 97–126 (2003)

    work page 2003

  13. [13]

    De Bouard, A

    A. De Bouard, A. Debussche,, Weak and strong order of convergence of a semidiscrete schem e for the stochastic nonlinear Schr ¨ odinger equation , Appl. Math. Optim. 54, pp. 369–399 (2006)

    work page 2006

  14. [14]

    De Bouard, A

    A. De Bouard, A. Debussche, L. Di Menza, Theoretical and numerical aspects of stochastic nonlinear Schr¨ odinger equations, Monte Carlo Methods Appl., 7, pp. 55–63 (2001)

    work page 2001

  15. [15]

    Debussche, L

    A. Debussche, L. Di Menza, Numerical simulation of focusing stochastic nonlinear Sch r¨ odinger equations, Phys. D 162, pp.131–154 (2002)

    work page 2002

  16. [16]

    Faou, Geometric Numerical Integration and Schr ¨ odinger equations , European Math.Soc

    E. Faou, Geometric Numerical Integration and Schr ¨ odinger equations , European Math.Soc. Pub- lishing House, Zurich (2012)

    work page 2012

  17. [17]

    Y. Feng, G. Maierhofer, K. Schratz, Long-time error bounds of low-regularity integrators for n on- linear Schr ¨ odinger equations , Math. Comp., to appear (2024)

    work page 2024

  18. [18]

    Liu, A mass-preserving splitting scheme for the stochastic Schr ¨ odinger equation with multiplica- tive noise , IMA J

    J. Liu, A mass-preserving splitting scheme for the stochastic Schr ¨ odinger equation with multiplica- tive noise , IMA J. Numer. Anal. 33, pp. 1469–1479 (2013)

    work page 2013

  19. [19]

    Liu, Order of convergence of splitting schemes for both determin istic and stochastic nonlinear Schr¨ odinger equations , SIAM J

    J. Liu, Order of convergence of splitting schemes for both determin istic and stochastic nonlinear Schr¨ odinger equations , SIAM J. Numer. Anal. 51, pp. 1911–1932 (2013)

    work page 1911

  20. [20]

    Hairer, C

    E. Hairer, C. Lubich, G. W anner, Geometric Numerical Integration: Structure-Preserving A lgo- rithms for Ordinary Differential Equations , Springer, Berlin (2002). 18 STEF ANO DI GIOV ACCHINO AND KATHARINA SCHRATZ

    work page 2002

  21. [21]

    Hairer, S

    E. Hairer, S. P. Norsett, G. W anner, Solving Ordinary Differential Equations I. Nonstiff Problem s. Second edition. Springer, Berlin (1993)

    work page 1993

  22. [22]

    Higham, P.E

    D.J. Higham, P.E. Kloeden, An Introduction to The Numerical Simulation of Stochastic D ifferential Equations, SIAM (2021)

    work page 2021

  23. [23]

    J. Hong, L. Sun, Symplectic Integration of Stochastic Hamiltonian Systems , Springer (2022)

    work page 2022

  24. [24]

    J. Hong, X. W ang, Invariant Measures for Stochastic Nonlinear Schr¨ odinger Equations. Numerical Approximations and Symplectic Structures , Springer (2019)

    work page 2019

  25. [25]

    Jiang, L

    S. Jiang, L. W ag, J. Hong, Stochastic multi-symplectic integrator for stochastic no nlinear Schr¨ odinger equation , Commun. Comput. Phys.14, pp. 393–411 (2013)

    work page 2013

  26. [26]

    Lord, C.E

    G.J. Lord, C.E. Powell, T. Shardlow, An Introduction to Computational Stochastic PDEs , Cam- bridge University Press (2014)

    work page 2014

  27. [27]

    C. M. Mora, R. Rebolledo, Regularity of solutions to linear stochastic Schr ¨ odinger equations, Infin. Dimens. Anal. Quantum Probab. Relat. Top., 10, pp. 237–259 ( 2007)

    work page 2007

  28. [28]

    Stochastic numerical modelling for sustaina ble innovation

    A. Ostermann, K. Schratz, Low Regularity Exponential-Type Integrators for semilinear Schr ¨ odinger Equations, Found. Comp. Math. 18, pp. 731–755 (2018). Acknowledgments SDG is member of the INdAM Research group GNCS. This work is partia lly supported by GNCS-INDAM project. The research is partially supported by PR IN-MUR 2022 project 20229P2HEA “Stochast...

    work page 2018