Long time decay and asymptotics for the complex mKdV equation
Gavin Stewart
TL;DR
The paper analyzes the long-time behavior of the complex mKdV equation with small localized data, proving that its asymptotics match those known for the real case. A key innovation is decomposing the solution as $u=S+w$, with $S$ a self-similar profile whose mean tracks the evolving zero-mode, and $w$ a remainder whose low-frequency content decays faster thanks to cancellations in $|S|^2\partial_x S$. The authors develop a robust, largely algebra-structure–independent approach via the space-time resonance method, obtaining precise region-wise asymptotics: rapid decay to the right, modified scattering on the left, and self-similar behavior in the core region, all governed by a Painlevé II–type self-similar equation. They further establish weighted energy estimates and a careful Fourier-space analysis to close a bootstrap and prove global existence for small data. The techniques are adaptable to other complex mKdV–type equations and highlight a robust framework for Painlevé-type long-time asymptotics beyond complete integrability.
Abstract
We study the asymptotics of the complex modified Korteweg-de Vries equation $\partial_t u + \partial_x^3 u = -|u|^2 \partial_x u$, which can be used to model vortex filament dynamics. In the real-valued case, it is known that solutions with small, localized initial data exhibit modified scattering for $|x| \geq t^{1/3}$ and behave self-similarly for $|x| \leq t^{1/3}$. We prove that the same asymptotics hold for complex mKdV. The major difficulty in the complex case is that the nonlinearity cannot be expressed as a derivative, which makes the low-frequency dynamics harder to control. To overcome this difficulty, we introduce the decomposition $u = S + w$, where $S$ is a self-similar solution with the same mean as $u$ and $w$ is a remainder that has better decay. By using the explicit expression for $S$, we are able to get better low-frequency behavior for $u$ than we could from dispersive estimates alone. An advantage of our method is its robustness: It does not depend on the precise algebraic structure of the equation, and as such can be more readily adapted to other contexts.
