Fractional Nonlinear Schrödinger Equation (fNLSE)

• The fNLSE is a nonlinear dispersive PDE where the classical Laplacian is replaced by its fractional power, introducing nonlocal, anomalous dispersion effects.
• It models complex phenomena such as Lévy-type dispersion, soliton dynamics, blow-up behavior, and ground states in optical, quantum, and BEC contexts.
• Advanced numerical methods, including spectral and finite-difference schemes, are employed to capture continuum limits and ensure stability in simulations.

The fractional nonlinear Schrödinger equation (fNLSE) is a fundamental nonlinear dispersive PDE in which the standard Laplacian is replaced by its fractional power, modeling nonlocal dispersion, with applications spanning mathematical physics, optics, Bose–Einstein condensates, and quantum many-body systems. The fractional Laplacian (−Δ)^{α/2}, with α∈(0,2){1}, introduces anomalous (often Lévy-type) dispersive effects and substantially alters the analytical and qualitative landscape relative to the classical NLSE. Rigorous studies address well-posedness, continuum limits from discrete models, ground-state and blow-up phenomena, and high-precision numerical simulation in both deterministic and stochastic frameworks.

1. Mathematical Formulation and Core Models

The standard fNLSE on ℝ^d takes the form

$$\partial_t u + i(-\Delta)^{\alpha/2}u = i\lambda|u|^{p-1}u, \qquad u(x,0)=u_0(x), \qquad x\in\mathbb{R}^d,$$

where:

• α∈(0,2){1} is the dispersion exponent (fractional order),
• (−Δ)^{α/2} is the fractional Laplacian, defined via the Fourier transform as ℱ^{-1}[ |ξ|^α ℱu(ξ)](x),
• p>1 governs the nonlinearity (with scaling/criticality distinguished by p and α),
• λ=±1 determines focusing (−1) or defocusing (+1) dynamics.

Discrete variants, essential for both analytic continuum-limits and numerical implementation, replace the fractional Laplacian by its discrete spectral or long-range difference realization: $$i\partial_t u_h(x,t)= (−\Delta_h)^{\alpha/2}u_h + \lambda|u_h|^{p-1}u_h, \quad x\in h\mathbb{Z}^d,$$ with (−Δ_h)^{α/2} defined by discrete Fourier multipliers on ℓ^2(h\mathbb{Z}^d) (Wang, 18 Jan 2025).

Generalizations include critical-power nonlinearities, spatially or temporally varying nonlocality, stochastic perturbations, and coupled/multicomponent systems with fractional group-velocity dispersion or Riesz derivatives (Zangmo et al., 2024).

2. Well-Posedness, Continuum Limit, and Stability

Well-posedness of the fNLSE is established in suitable Sobolev spaces H^{α/2}(ℝ^d), with local existence typically for s>s_c=(d/2−α/(p−1)), and global existence in subcritical or defocusing regimes. For the discrete fNLSE, global well-posedness in ℓ^2(h\mathbb{Z}^d) follows for all data.

A fundamental result is the rigorous continuum limit: solutions u_h to the discrete problem, under sampling and piecewise-linear interpolation, converge strongly in L^2(ℝ^d) to solutions u of the continuum fNLSE as the lattice spacing h→0. Precise rates are given: for d=3, p∈[3,5), and suitable α,

$$\|p_h u_h(t) - u(t)\|_{L^2(\mathbb{R}^3)} \leq C_1 h^{\alpha/(2+\alpha)} e^{C_2 t},$$

with constants C_1,C_2>0 (Wang, 18 Jan 2025). The proof relies on uniform-in-h Strichartz estimates and fine analysis of oscillatory-integral kernels using Newton polyhedron techniques. Analogous continuum-limit results hold in d=2, though with exponents and sharp constants depending on α and the criticality window (Choi et al., 2022).

These results justify the accuracy and stability of spectral and finite-difference numerical schemes in simulating the fNLSE for a broad range of parameters.

3. Discrete Fractional Laplacians and Nonlocality

The discrete fractional Laplacian is fundamental to both rigorous analysis and physical modeling, realized in both one and higher dimensions as a nonlocal coupling operator. In 1D,

$$(−\Delta)^\alpha \psi_n = \sum_{m\neq n}K^\alpha(n−m)\bigl[\psi_n−\psi_m\bigr],$$

where K^\alpha(m) ~ |m|^{−(1+2\alpha)} for |m|→∞, reflecting power-law decay and infinite-range coupling. The kernel is intimately related to Gamma functions and can be written in analogous forms in d=2 with Bessel-function representations (Molina, 2019, Molina, 2020).

The fractional exponent serves as a tunable parameter interpolating from local (α≈1, standard Laplacian) to increasingly nonlocal (α→0^+) dynamics, leading to emergent phenomena such as band flattening, quasi-degenerate modes, slower ballistic transport, and lower self-trapping thresholds.

In high dimensions, the technical complexity of establishing uniform dispersive/Strichartz estimates is handled via detailed stationary-phase analysis and the Newton polyhedron method, crucial for the convergence and stability results aforementioned (Wang, 18 Jan 2025, Choi et al., 2022).

4. Ground States, Blow-Up, and Criticality

In focusing fNLSE, ground-state (positive least-energy) solutions are constructed variationally for subcritical and critical nonlinearities in general potentials, without requiring monotonicity or Ambrosetti–Rabinowitz conditions (Jin et al., 2016). The critical exponent and the algebraic decay of ground-state profiles—as |x|^{−(N+2s)} in ℝ^N—reflect the nonlocality and lack of exponential localization characteristic of integer-Laplacian problems.

Blow-up phenomena in the fNLSE are governed by the interplay of nonlinearity strength (p, α), dimension d, and initial data. Explicit sufficient blow-up criteria for nonradial, irregular data in L^2 or energy-subcritical settings are obtained via localized virial-type estimates:

• If along a solution the indicator K(u(t)) ≤ −δ < 0 for all t, either finite-time blow-up or norm divergence at infinity is ensured (Dinh, 2018).
• Mass-critical and intercritical regimes, as well as explicit thresholds in terms of ground states Q and energies, are precisely characterized (Dinh, 2018, Dinh, 2018).
• Concentration-compactness methods reveal that minimal-mass blow-up concentrates the ground-state profile Q (Dinh, 2018).

Numerically, the nature of blow-up, global existence, and semiclassical dynamics in one dimension are analyzed in detail. The transition through critical regimes, the formation of algebraic soliton tails, and the structure of dispersive shocks is documented (Klein et al., 2014).

5. Solitary Waves and Dynamics

Solitary and multi-soliton solutions of the fNLSE with cubic or more general nonlinearity are central to both theory and numerical simulation. Analytical and spectral numerical methods (e.g., Petviashvili iteration, Newton–Krylov solvers) produce ground-state and moving soliton profiles, which exhibit sharper localization and algebraic tails as the fractional order decreases (Bayindir et al., 2021, Durán et al., 12 Jul 2025). Direct numerical integration confirms:

• Soliton solution width shrinks with decreasing α, amplitude increases, and solutions develop power-law decay |x|^{−(2s+1)} at infinity in 1D (Durán et al., 12 Jul 2025).
• Stability is modulated through the Vakhitov–Kolokolov criterion; dynamical simulations show that solitons split, spread, or coalesce, particularly under perturbations or external potentials.
• Collisions in fractional coupled systems (e.g., in bimodal fiber cavities) can produce rebounds, mergers, splitting into breathers, and elastic or inelastic passage, controlled by the Lévy index and group-velocity mismatch (Zangmo et al., 2024).
• In 2D and 3D, solitary wave phenomena, including ground states and their spectral stability, remain only partially classified due to the complexity induced by nonlocality.

Asymptotic dynamics and resolution properties demonstrate non-integrable soliton interactions, long-lived breathers, and inelastic deformations mediated by nonlocal dispersion (Durán et al., 1 Aug 2025).

6. Numerical Methods and Structure-Preserving Algorithms

Accurate simulation of fNLSE dynamics is achieved via Fourier spectral, structure-preserving, and high-order time-integration schemes:

• Energy-conserving methods based on the scalar auxiliary variable (SAV) approach guarantee exact preservation of Hamiltonian invariants at the discrete level, with summation-by-parts and Crank–Nicolson time stepping yielding second-order temporal and spectral spatial convergence (Fu et al., 2019).
• Fully discrete L1/CN schemes for time-fractional (Caputo) nonlinear Schrödinger dynamics in 2D achieve unconditional stability and demonstrate first-order (in time) and second-order (in space) convergence, validated via manufactured solutions and nonsmooth initial data (Ma et al., 14 Apr 2025).
• Petviashvili iteration, FFT-based spectral derivatives, and explicit Runge–Kutta methods are effective for constructing and propagating solitary waves in diverse potential landscapes (Bayindir et al., 2021).
• Discretizations for lattice models with power-law coupling are used both for analysis of the continuum limit and for simulating self-trapping and dynamical regimes in optical and condensed-matter contexts (Molina, 2019, Molina, 2020).

Seminal works validate the justification and reliability of finite-difference and pseudospectral methods via rigorous L^2-convergence as mesh-size h→0 in energy-subcritical regimes (Wang, 18 Jan 2025, Choi et al., 2022).

7. Extensions: Stochastic, Generalized, and Topological fNLSEs

Recent advances consider stochastic fNLSEs and more generalized nonlocal or time-fractional variants:

• Slow-fast stochastic fNLSEs, governed by a fractional dispersive generator and nonlinearities with polynomial growth, admit averaging principles in the strong sense. Upon scale separation, the slow component converges in mean square to an effective averaged stochastic PDE, with technical tools including vanishing-viscosity regularization, ergodicity, and Khasminskii-type time discretization (Mohan et al., 12 May 2026).
• Space–time fNLSEs with Jumarie-modified Riemann–Liouville derivatives allow for explicit analytical solution families, including Mittag–Leffler plane waves, Riccati-type bright/dark solitons with memory-dependent transitions, and explicit links to fractional Fokker–Planck and Q-Gaussian statistics (Najafi et al., 25 May 2025).
• In periodic and topological media, the fNLSE with honeycomb potential supports effective Dirac equations with spatially modulated mass, and rigorous error bounds demonstrate the reduction of wavepacket dynamics to nonlinear two-component Dirac fields (Xie et al., 2020).

These directions integrate nonlocal dispersive analysis with stochastic modeling and topological band theory, broadening the scope and applicability of fNLSE dynamics.

---

References: (Wang, 18 Jan 2025, Choi et al., 2022, Molina, 2020, Molina, 2019, Jin et al., 2016, Ma et al., 14 Apr 2025, Fu et al., 2019, Klein et al., 2014, Dinh, 2018, Dinh, 2018, Zangmo et al., 2024, Durán et al., 12 Jul 2025, Durán et al., 1 Aug 2025, Bayindir et al., 2021, Najafi et al., 25 May 2025, Xie et al., 2020).