Lattice doubling and the linear stability of solitary waves in the complementary generalized ABS model
Year: 2026
Type: article
Abstract: The linear stability of solitary waves of the nonlinear Dirac equation in $1+1$ dimensions has resisted direct numerical computation because the discretized linearized operator produces eigenvalues that do not survive the continuum limit. We identify the mechanism as species doubling. Linearizing the complementary generalized Alexeeva--Barashenkov--Saxena (cABS) model about its exact solitary waves, we compute the spectrum of the stability operator with four lattice derivatives: the symmetric (naive) difference, the Wilson derivative, the SLAC derivative, and the chiral disordered derivative of Jacobs (1983). The naive derivative produces real unstable eigenvalues that are independent of the lattice spacing and carry their entire Fourier weight at the edge of the Brillouin zone; this doubler sector is the continuum stability operator evaluated at the spatially reflected profile $\psi_0(-x)$, which is not a stationary solution, so that its spectrum is that of a definite, wrong problem rather than numerical noise. The chiral disordered derivative, which lifts the doublers by a chirality-preserving, non-Hermitian mechanism, renders the stability operator non-Hermitian and is unusable for a real-time problem, in line with the symmetry-class criterion of arXiv:2611.05395. The Wilson and SLAC derivatives agree as $a\to0$ and isolate a third signature, complex quartets in the overlap of the two continuum branches with real parts proportional to $1/L$, which are finite-volume Krein collisions; a box-size sweep at every spectrally stable point confirms that none survives the infinite-volume limit. With these signatures separated, the physical spectrum is unambiguous and shows two instabilities. A real unstable eigenvalue is present wherever $dQ/d\omega>0$, equivalently wherever $E/Q$ decreases with $\omega$ (the two are the same statement, by $E'(\omega)=\omega Q'(\omega)$), which confirms the surmise of Khare, Cooper, Dawson and Saxena, identifies it with the Vakhitov--Kolokolov sign condition, and reconciles it with their non-relativistic result. The sign condition is sufficient but not necessary: at $q=1.1$ and small $\omega$, where $dQ/d\omega<0$, a complex quartet appears for $\kappa$ from about $1.4$ to beyond $2.4$, with a growth rate that rises from $0.06$ to $0.55$ across that range and a frequency between $1.3$ and $2.4$, converged in lattice spacing and box size up to $L=320$. It is an oscillatory instability confined to the window $1.05<q<1.15$, with growth rates up to $0.75$ at $q=1.075$: it marks the region where the scalar term nearly cancels the vector term. Solitary waves with $E/Q>1$ but no unstable eigenvalue exist, so $E/Q<1$ is a binding criterion, not a stability criterion. Below a threshold $\kappa_c(q)$, computable from the closed-form charge, equal to $1+0.49\sqrt{\epsilon}+O(\epsilon\ln\epsilon)$ near $q=1$, $\epsilon=(q-1)/q$, and to the non-relativistic value $2$ for $q\ge7$, every solitary wave is spectrally stable for all $0<\omega<m$.
Authors: Laurence A Jacobs, Avinash Khare
Language: English
Topic: Nonlinear Waves and Solitons
Subfield: Statistical and Nonlinear Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7235281255
