Linearized Soliton Perturbation Theory

• Linearized Soliton Perturbation Theory (LSPT) is a family of methods that tracks how coherent structures like solitons respond to weak perturbations by expanding around a nonlinear state.
• LSPT employs various techniques—direct linearization, inverse-scattering, and collective-coordinate reductions—to derive reduced evolution equations and manage secular growth.
• The approach finds applications in systems such as nonlinear Schrödinger, Kadomtsev–Petviashvili, and water-wave models, enabling effective stability analysis of soliton behavior.

Searching arXiv for the cited papers and closely related LSPT terminology to ground the article in current arXiv records. Linearized Soliton Perturbation Theory (LSPT) is best understood, in the literature considered here, as a family of perturbative methods for tracking how an exact soliton, line soliton, periodic nonlinear wave, or other coherent structure responds to weak forcing, weak background fields, slow parameter variation, or small changes of initial data. The common mechanism is to expand about a distinguished nonlinear state, isolate the linearized operator and its neutral, discrete, resonant, or continuous modes, and then derive reduced evolution equations by solvability, orthogonality, inverse-scattering, or matched-asymptotic arguments. Across nonlinear Schrödinger, Kadomtsev–Petviashvili, Klein–Gordon, Schrödinger–Poisson, sigma-model, and water-wave settings, the resulting theories range from direct inhomogeneous linearization around a soliton profile to exact spectral linearization in scattering variables, and from collective-coordinate reductions to fixed-background mode expansions (1306.04445, Boiti et al., 2012, Zagorac et al., 2021).

1. Methodological scope

A basic LSPT template is an expansion of the field around a coherent background with a small parameter. In the extended nonlinear Schrödinger setting one writes $\psi=\psi_0+\psi_1$, where $\psi_1$ is a small correction to the standard $\sech$ profile $\psi_0$ (1306.04445). In the KP-II modulation problem one uses

$$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$

with $x$ retained as a fast variable (Han et al., 3 Mar 2025). Near a $\mathcal{PT}$-symmetric phase-transition point, the localized wave is expanded as

$$\psi(x,z)=\left(\epsilon u_1(x,Z)+\epsilon^2u_2+\epsilon^3u_3+\cdots\right)e^{-i\mu_0 z}, \qquad Z=\epsilon z,$$

so that the perturbation problem is organized by the Jordan-chain structure of the exceptional point (Nixon et al., 2015).

This range of ansätze suggests that LSPT is not a single formalism but an umbrella for several closely related procedures. One branch emphasizes direct linearization around a localized profile and the removal of secular growth by orthogonality conditions. A second branch uses inverse-scattering variables and tracks perturbations of discrete eigenvalues, norming constants, and reflection data. A third branch keeps only symmetry-generated or low-lying collective coordinates and derives effective finite-dimensional dynamics in a weak background. A fourth branch treats periodic or multisoliton backgrounds, where the relevant “linear modes” are Bloch, Floquet, Lamé, or dressed scattering states rather than decaying eigenfunctions (Lashkin, 2021, Mullyadzhanov et al., 2020, Harland, 2016).

2. Linearization, neutral modes, and solvability

In direct LSPT, the central object is the inhomogeneous linearized equation around the unperturbed coherent structure. For the extended NLSE with inhomogeneous second-order diffraction and pseudo-stimulated Raman scattering, the first-order correction $\psi_1$ satisfies

$$q_0 \frac{d^2\psi_1}{d\xi^2}+\left(6\alpha \psi_0^2-2\Omega\right)\psi_1 = - q' \xi \frac{d^2\psi_0}{d\xi^2} - q' \frac{d\psi_0}{d\xi} - \mu \psi_0 \frac{d(\psi_0^2)}{d\xi},$$

which is explicitly the linearized inhomogeneous equation around the $\psi_1$0 soliton (1306.04445). The paper shows that localization of the first-order correction requires the specific equilibrium value

$\psi_1$1

and that if $\psi_1$2, the correction diverges at infinity. This is a standard LSPT pattern: boundedness of the first-order correction is equivalent to a balance or solvability condition.

For KP line solitons, the same structure appears in a Whitham-type setting. After promoting the exact line-soliton parameters $\psi_1$3 to slowly varying fields, the $\psi_1$4 correction satisfies

$\psi_1$5

with $\psi_1$6. Orthogonality of the right-hand side to $\psi_1$7 yields the closed modulation system for $\psi_1$8, namely

$\psi_1$9

In this formulation, rarefaction solutions generate “parabolic solitons,” while shocks are regularized by resonant generation of new line solitons (Han et al., 3 Mar 2025).

Near non-Hermitian exceptional points, the neutral-mode structure is more singular. For localized waves in $\sech$0-symmetric potentials, the shifted operator satisfies

$\sech$1

so the linearized theory involves a Jordan chain rather than a semisimple kernel. Projection onto the adjoint nullspace yields the reduced amplitude equation

$\sech$2

with

$\sech$3

Here the solvability mechanism is still recognizably LSPT, but the reduced dynamics is second order in the slow variable because of the exceptional-point Jordan structure (Nixon et al., 2015).

These examples support a common interpretation: the decisive technical step in LSPT is not merely writing a small perturbation expansion, but identifying which components of the forcing lie in the range of the linearized operator and which must be removed or renormalized by modulation equations.

3. Inverse-scattering and spectral formulations

A major branch of LSPT works in spectral variables rather than physical-space collective coordinates. For the Davydova–Lashkin–Fokas–Lenells equation, the perturbed problem

$\sech$4

is treated by inverse scattering, with discrete scattering data $\sech$5 for solitons and continuous data $\sech$6 or $\sech$7 for radiation. The exact perturbation equations include

$\sech$8

together with an evolution law for the norming constants. The paper explicitly states the standard approximation: if $\sech$9 is small, substitute the unperturbed $\psi_0$0-soliton Jost functions into the right-hand side of the exact evolution equations. In LSPT language, this is the leading-order projection of the perturbation onto the discrete neutral manifold and the continuous radiation modes (Lashkin, 2021).

For long-lived deep-water bi-solitons, the same spectral idea is applied in a hybrid analytical-numerical way. The exact or approximate water-wave evolution is embedded into a perturbed focusing NLS flow

$\psi_0$1

and the perturbation of each Zakharov–Shabat eigenvalue obeys

$\psi_0$2

The coherent structure is then interpreted as a two-soliton core plus continuous-spectrum radiation, with periodic energy and momentum exchange between the discrete and continuous parts (Gelash et al., 2023).

The most explicit first-order scattering-data perturbation theory in the corpus is developed for the focusing 1D NLSE with a box-shaped initial condition. The paper linearizes the full scattering map

$\psi_0$3

and derives the eigenvalue sensitivity formula

$\psi_0$4

It also gives

$\psi_0$5

so the perturbation theory closes simultaneously on discrete eigenvalues, norming constants, and reflection data. In the box problem this leads to closed-form sensitivities for soliton amplitude, velocity, phase, and position, and to the “virtual soliton eigenvalue” construction for noise-induced soliton emergence or disappearance (Mullyadzhanov et al., 2020).

An exact background-adapted version of the same idea exists for KPII on an $\psi_0$6-soliton background. There the direct and inverse problems are formulated for the heat operator with potential $\psi_0$7, and the nonlinear initial-value problem is “linearized” in the IST sense because the continuous spectral data evolve by

$\psi_0$8

This is not small-amplitude linearization of the PDE around $\psi_0$9, but exact spectral linearization relative to a nontrivial multisoliton background (1212.07259).

4. Background-sensitive extensions

A recurring lesson in the literature is that linearization around the soliton core alone may be insufficient. The theory of dark-soliton perturbations makes this explicit. For the perturbed defocusing NLS

$$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$0

the inner perturbation expansion around the dark-soliton core produces first-order corrections that do not decay at infinity. The resulting mismatch with the outer background generates a propagating shelf, and the outer linearization gives

$$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$1

with long-wave shelf speed

$$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$2

The paper’s central methodological claim is that any consistent perturbation theory for dark solitons must include this shelf, since it contributes nontrivially to conservation-law balances and hence to the parameter evolution itself (Ablowitz et al., 2010).

A related extension appears when the perturbation is not a forcing of the soliton equation but a weak external field acting on soliton moduli. For sigma models whose target is a symmetric space, the derived weak-background Lagrangian shows that the background field modifies the usual moduli-space approximation in two ways: by introducing a potential energy, and by inducing a Kaluza–Klein metric on the moduli space. In the Skyrme-model application, quantization of the one-soliton dynamics reproduces the leading pion–nucleon term in the chiral effective Lagrangian (Harland, 2016). This suggests a collective-coordinate version of LSPT in which the perturbation lives in the background field and the retained dynamical variables are soliton zero modes.

For self-gravitating Schrödinger–Poisson solitons, the perturbation theory is again background-sensitive. One first solves the linear Schrödinger eigenproblem in the frozen ground-state soliton potential,

$$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$3

and then includes backreaction through a time-dependent potential perturbation. The mode amplitudes satisfy

$$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$4

This is not a full linearization of the coupled Schrödinger–Poisson system about the soliton, but rather a fixed-background mode expansion plus perturbative self-gravity correction (Zagorac et al., 2021).

The wave-dark-matter soliton literature pushes this logic further by emphasizing that a collective excitation of a self-gravitating soliton necessarily mixes positive- and negative-energy components. With

$$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$5

the linearized collective equations become

$$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$6

and the natural Hermitian structure involves the mixed quantity $$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$7, not the naive $$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$8 norm of $$\phi(x,y,t)=\phi^{(0)}(\xi,Y,T)+\epsilon \phi^{(1)}(\xi,Y,T)+O(\epsilon^2), \qquad Y=\epsilon y,\quad T=\epsilon t,$$9 or $x$0 alone. A constrained variational principle then predicts the $x$1 collective-mode frequencies, with $x$2 requiring the strongest negative-energy coupling and $x$3 almost none (Chiueh et al., 2022).

5. Periodic, lattice, and line-soliton backgrounds

Several recent works extend LSPT beyond isolated localized pulses to periodic or transversely extended nonlinear backgrounds. For the KdV equation, the cnoidal wave

$x$4

is treated as a periodic soliton lattice, and perturbations are organized by the Lamé spectral problem

$x$5

In this setting, continuous-spectrum perturbations correspond to Bloch waves in the allowed bands, while localized defects correspond to spectral data in forbidden bands. The one-defect solution

$x$6

satisfies

$x$7

so the perturbation is best interpreted as a localized lattice dislocation carrying a net phase shift $x$8, not as a decaying additive field on a fixed periodic background (Kuznetsov, 31 Mar 2026).

For KP-II line solitons, the modulation variables are the local amplitude $x$9 and local phase shift $\mathcal{PT}$0 along the crest. The nonlinear stability theorem proves that the amplitude of the line soliton converges to that of the line soliton at initial time whereas jumps of the local phase shift of the crest propagate in a finite speed toward $\mathcal{PT}$1. The local amplitude and the phase shift are described by a system of 1D wave equations with diffraction terms, and after diagonalization the asymptotics reduce to coupled Burgers equations moving along the rays $\mathcal{PT}$2 (Mizumachi, 2013). This is a rigorous transverse-modulation version of LSPT in which the neutral directions form a low-frequency resonant bundle over the transverse Fourier parameter.

The KPII half-soliton and $\mathcal{PT}$3-shaped-data problem leads to a complementary modulation picture. There the $\mathcal{PT}$4-system

$\mathcal{PT}$5

produces rarefaction sectors interpreted as parabolic solitons and shock singularities interpreted as resonant generation of new line solitons (Han et al., 3 Mar 2025). This suggests that on transversely modulated soliton backgrounds, the output of linearized solvability may be a nonlinear first-order PDE in soliton parameters rather than an ODE for a few collective coordinates.

6. Zero modes, resonances, and structural obstructions

One common misconception is that the main difficulty in LSPT is only the computation of a first-order correction. Several papers show that the deeper issue is the treatment of structural obstructions.

At the quantum level, translational zero modes obstruct ordinary perturbation theory for soliton states because the center-of-mass motion has continuous spectrum. The proposed remedy is to restrict first to fixed total momentum, and in particular to the $\mathcal{PT}$6 sector for the soliton ground state. In the shifted problem one imposes

$\mathcal{PT}$7

order by order, which fixes the dangerous zero-mode dependence of the wavefunctional and restores perturbative invertibility (Evslin, 2020).

At the PDE level, threshold resonances can play a comparable role. For the 1D focusing cubic Klein–Gordon equation, linearization about the static soliton $\mathcal{PT}$8 yields

$\mathcal{PT}$9

with one negative eigenmode and an even threshold resonance

$$\psi(x,z)=\left(\epsilon u_1(x,Z)+\epsilon^2u_2+\epsilon^3u_3+\cdots\right)e^{-i\mu_0 z}, \qquad Z=\epsilon z,$$0

The paper shows that the threshold resonance of the linearized operator produces a one-dimensional space of slowly decaying Klein–Gordon waves, relative to local norms, and that the key quadratic resonance coefficient does not vanish: $$\psi(x,z)=\left(\epsilon u_1(x,Z)+\epsilon^2u_2+\epsilon^3u_3+\cdots\right)e^{-i\mu_0 z}, \qquad Z=\epsilon z,$$1 The main obstruction to full asymptotic stability on the center-stable manifold is therefore a small divisor in a quadratic source term of the perturbation equation, and the resulting stability theorem gives decay only up to exponential time scales (Luhrmann et al., 2023).

A different obstruction appears in complex potentials. For generalized nonlinear Schrödinger equations with non-$$\psi(x,z)=\left(\epsilon u_1(x,Z)+\epsilon^2u_2+\epsilon^3u_3+\cdots\right)e^{-i\mu_0 z}, \qquad Z=\epsilon z,$$2-symmetric complex potentials, one would normally expect no continuous nonlinear families bifurcating from linear discrete modes because straightforward perturbation of the complex stationary equation typically generates infinitely many solvability conditions. The special potentials

$$\psi(x,z)=\left(\epsilon u_1(x,Z)+\epsilon^2u_2+\epsilon^3u_3+\cdots\right)e^{-i\mu_0 z}, \qquad Z=\epsilon z,$$3

avoid this by admitting a constant of motion and a reduction to a real amplitude equation, so that continuous families of solitons always bifurcate out from linear discrete modes in these non-$$\psi(x,z)=\left(\epsilon u_1(x,Z)+\epsilon^2u_2+\epsilon^3u_3+\cdots\right)e^{-i\mu_0 z}, \qquad Z=\epsilon z,$$4-symmetric complex potentials (Nixon et al., 2015). This is a reminder that what counts as a “generic obstruction” in LSPT can disappear when the background has hidden structure.

Finally, several spectral formulations emphasize that radiation is not a negligible correction to the soliton sector. The deep-water bi-soliton work interprets the long-lived coherent structure as a balance between a dominant solitonic part and a portion of continuous spectrum radiation, with periodic exchange of energy and momentum between them (Gelash et al., 2023). The box-potential IST perturbation theory reaches a related conclusion from the opposite direction: one should perturb the full scattering portrait $$\psi(x,z)=\left(\epsilon u_1(x,Z)+\epsilon^2u_2+\epsilon^3u_3+\cdots\right)e^{-i\mu_0 z}, \qquad Z=\epsilon z,$$5, not only the discrete eigenvalues, because the continuous spectrum participates in parameter shifts, noise statistics, and threshold crossing via virtual eigenvalues (Mullyadzhanov et al., 2020).

Taken together, these results suggest that LSPT is most successful when it is treated as a structured perturbation theory of the full soliton background—its neutral modes, generalized kernels, continuous spectrum, and conservation constraints—rather than as a bare Taylor expansion around a single profile.