Strang Splitting Methods for a quasilinear Schrödinger equation - Convergence, Instability and Dynamics

Jianfeng Lu, Jeremy L. Marzuola

Abstract

This paper investigates the numerical approximation of a class of quasilinear Schrödinger equations (QLS) using the Strang splitting method. By adapting Lie theoretic approaches and Sobolev-based well-posedness results, we prove that the scheme achieves second-order convergence for small initial data over finite time intervals. We complement this analysis with numerical simulations of a 1D model related to superfluid thin films and relativistic laser pulses. Our results verify the theoretical convergence rates and identify a critical amplitude threshold for stability, beyond which solutions exhibit singularity formation.

Introduction

The Nonlinear Schrödinger (NLS) equation is a fundamental model in physics, describing wave propagation in media ranging from fiber optics to water waves. While "semilinear" versions—where the nonlinearity depends only on the solution's magnitude—are well-understood, many physical systems require "quasilinear" models. In these cases, the nonlinearity involves derivatives of the solution itself, significantly increasing the mathematical complexity and numerical stiffness.

Quasilinear Schrödinger equations are particularly relevant for modeling superfluid thin films and the propagation of ultra-short pulse lasers in relativistic regimes. This study focuses on the Strang splitting method, a popular time-discretization algorithm, to determine its efficacy in handling the structural challenges of QLS.

Methodology

The Strang Splitting Scheme

The core of the approach is the decomposition of the complex evolution of the PDE into two simpler, alternating steps:

  1. The Linear Step: A dispersive step solved exactly in frequency space.
  2. The Nonlinear Step: A step that, for the specific models studied, preserves the local amplitude $|u|^2$ of the wave.

By splitting the operator, the authors create a symmetric scheme that is both computationally efficient and capable of preserving the physical structures of the system, such as mass and energy.

Spatial Discretization and Regularization

For spatial resolution, a Fourier pseudo-spectral method is employed. This uses Fast Fourier Transforms (FFTs) to achieve high accuracy on periodic domains. To maintain stability in the presence of high-frequency "stiffness," the authors implement a spectrum truncation filter. This filter zeros out negligible Fourier coefficients, preventing numerical noise from destabilizing the evolution during the nonlinear phase.

Theoretical Convergence

The theoretical contribution of the paper involves proving that the numerical error scales with the square of the time-step ($τ^2$). This is achieved by adapting the Lie theoretic framework established by Lubich for NLS equations to the more demanding quasilinear setting. The proof holds for small initial data in high-regularity Sobolev spaces $H^s$, ensuring that the numerical approximation remains close to the true solution over finite time intervals.

Numerical Experiments and Dynamics

To test the limits of the theory, the authors simulated a 1D modified superfluid thin-film equation using two primary types of initial conditions: symmetric Gaussian profiles and plane wave trains.

Order Verification

Numerical tests confirmed the predicted second-order convergence in both $L^2$ and $H^1$ norms. Doubling the temporal resolution (halving $τ$) consistently resulted in a fourfold reduction in error, validating the Strang splitting approach for this class of equations.

The Stability Threshold and Blow-up

A significant finding of the numerical study is the identification of a critical stability threshold. Through linear stability analysis and direct simulation, the authors found that solutions with amplitudes below approximately $0.707$ ($\sqrt{2}/2$) remain stable.

However, when the initial data exceeds this threshold, the system enters a "focusing" regime. The solution does not merely oscillate; it forms a narrow, high-amplitude spike—a singularity. This "blow-up" behavior is a physical characteristic of the underlying equation rather than a failure of the numerical scheme. The Strang splitting method successfully captured the dynamics of these singularities, showing how the solution "squeezes" into an increasingly tall and narrow peak.

Discussion and Conclusion

This research demonstrates that Strang splitting is a robust and reliable tool for quasilinear Schrödinger equations, provided the initial data is within stable bounds. The method's ability to preserve mass and energy makes it ideal for long-term simulations of physical phenomena like laser pulse propagation.

Future Directions

While the convergence proof is rigorous for small data, extending this analysis to "large data" near the point of singularity remains an open challenge. Additionally, while the current study is one-dimensional, extending the numerical framework to 2D and 3D domains would allow for the exploration of more complex geometric effects and higher-dimensional blow-up patterns. The striking correlation between the plane-wave stability threshold and the behavior of Gaussian wave packets suggests a deeper mathematical link that warrants further formal investigation.

Summary updated

Abstract

We study the Strang splitting scheme for quasilinear Schrödinger equations. We establish the convergence of the scheme for solutions with small initial data. We analyze the linear instability of the numerical scheme, which explains the numerical blow-up of large data solutions and connects to analytical breakdown of regularity of solutions to quasilinear Schrödinger equations. Numerical tests are performed for a modified version of the superfluid thin film equation.
BibTeX
@inproceedings{lu2015strang,
  author = {Jianfeng Lu and Jeremy L. Marzuola},
  title = {{Strang Splitting Methods for a quasilinear Schrödinger equation - Convergence, Instability and Dynamics}},
  booktitle = {Communications},
  year = {2015},
  doi = {10.4310/cms.2015.v13.n5.a1},
}
Figure