A numerical method for the fractional Zakharov-Kuznetsov equation

Technical Summary: A Numerical Method for the Fractional Zakharov–Kuznetsov Equation

Problem Statement
This paper addresses the numerical approximation of the fractional Zakharov–Kuznetsov (fZK) equation posed on a two-dimensional periodic domain $\Omega = [-\pi, \pi]^2$. The equation is given by:
$$\partial_t u - \partial_{x_1}(-\Delta)^{\alpha/2}u + u\partial_{x_1}u = 0,$$
where $\alpha \in (0, 2]$. This model generalizes the classical Zakharov–Kuznetsov (ZK) equation ($\alpha=2$) and the higher-dimensional Benjamin–Ono–ZK model ($\alpha=1$) to include weaker fractional-dispersion regimes ($0 < \alpha < 1$). The fractional Laplacian $(-\Delta)^{\alpha/2}$ is defined spectrally via Fourier series. The system possesses fundamental conserved quantities: mass, momentum, and Hamiltonian energy. While analytical well-posedness has been studied for various $\alpha$, rigorous numerical methods with complete convergence analysis for the fZK equation have been largely unexplored.

Methodology
The authors propose a fully discrete Fourier spectral Galerkin (FSG) method combined with an integrating-factor fourth-order Runge–Kutta (IFRK4) time discretization.

1. Spatial Discretization (Semi-discrete FSG):

   • The solution is approximated in the space of trigonometric polynomials $X_N$ using the orthogonal Fourier projection $\Pi_N$.
   • The scheme is formulated to preserve the discrete analogues of the continuous mass, momentum, and Hamiltonian energy invariants.
   • Existence and uniqueness of the semi-discrete solution are established via standard ODE theory applied to the Fourier coefficient system.

2. Convergence Analysis (Spatial):

   • Strong Convergence: The authors prove that the semi-discrete approximations converge strongly to the unique local strong solution in $C([0, \bar{T}]; L^2_{per}(\Omega))$ for initial data in $H^s_{per}(\Omega)$ with $s \ge 2 + \alpha$. This relies on uniform Sobolev bounds, Rellich compactness, and the Arzelà–Ascoli theorem.
   • Optimal Error Estimates: A key technical innovation is the introduction of a modified projection operator $\tilde{\Pi}_N$ adapted to the fractional transport-dispersive operator. Unlike standard $L^2$ projections, this operator cancels the linear residual in the error equation.
   • Using refined adjoint estimates and periodic Kato–Ponce product/commutator estimates, the paper derives optimal spatial error estimates of order $O(N^{-r})$ for $r > 2 + \alpha$. For analytic solutions, exponential convergence is proven.

3. Time Discretization (Fully Discrete):

   • To handle the stiffness introduced by the fractional dispersive term (which yields eigenvalues of order $O(N^{1+\alpha})$), an integrating-factor approach is employed. The linear part is integrated exactly, while the nonlinear part is treated explicitly.
   • A fourth-order four-stage Runge–Kutta (IFRK4) scheme is applied to the transformed system.
   • Under a high-regularity nonlinear stability assumption (requiring the solution to remain in a bounded high Sobolev ball), the authors derive a fourth-order temporal error estimate, $O(\Delta t^4)$.

Key Results

• Conservation Properties: The semi-discrete scheme exactly conserves discrete mass, momentum, and energy. The fully discrete IFRK4 scheme does not strictly preserve these invariants but exhibits negligible drift in numerical experiments.
• Convergence Rates:
  • Spatial: Optimal algebraic convergence $O(N^{-r})$ for $r > 2 + \alpha$ and exponential convergence for analytic solutions.
  • Temporal: Fourth-order convergence $O(\Delta t^4)$.
  • Fully Discrete: The total error is bounded by $C(N^{-r} + \Delta t^4)$.
• Numerical Validation:
  • Spectral Accuracy: Experiments with the classical ZK equation ($\alpha=2$) using exact solitary wave solutions confirm exponential convergence, with errors decreasing by over nine orders of magnitude as $N$ increases from 16 to 256.
  • Long-Time Behavior: Simulations over $T=1000$ show that the fully discrete method maintains relative conservation errors of $O(10^{-14})$ for mass and $O(10^{-8})$ for momentum and energy.
  • Fractional Dynamics: Simulations of soliton interactions for various $\alpha \in \{0.5, 1, 1.5, 1.9, 2\}$ demonstrate the method's ability to capture the dependence of propagation speed and wave spreading on the fractional order.

Significance and Claims
The paper claims to provide the first complete numerical analysis for the fZK equation, bridging a critical gap in the literature where rigorous error analysis for such models was previously absent. The significance of the work lies in:

1. Structure Preservation: The development of a scheme that exactly preserves discrete conservation laws, ensuring long-term stability and physical fidelity.
2. Rigorous Analysis: The proof of existence, uniqueness, and strong convergence for the full fractional range $0 < \alpha \le 2$, extending well-posedness results to the numerical setting.
3. Technical Innovation: The use of a modified projection operator and refined adjoint estimates to achieve optimal error rates even in the presence of fractional transport-dispersive operators, a technique that remains valid for the challenging regime $0 < \alpha < 1$.
4. Practical Efficiency: The demonstration that the integrating-factor approach effectively removes the severe time-step restrictions associated with stiff dispersive terms, allowing for efficient fourth-order time integration.

The authors conclude that the proposed IFRK4-FSG framework is a robust and accurate tool for simulating nonlinear dispersive waves with fractional dispersion, validated by both theoretical error bounds and comprehensive numerical experiments.