[Paper Review] Lagrangian Controllability of the 1-D Korteweg-de Vries Equation

Ludovick Gagnon|arXiv (Cornell University)|Nov 9, 2016

Advanced Mathematical Physics Problems7 references3 citations

TL;DR

This paper establishes small-time Lagrangian controllability of the 1D Korteweg-de Vries (KdV) equation using N-soliton solutions. By constructing a smooth, localized solution with rapidly increasing soliton amplitudes, the authors show that particles initially in [0,L] can be driven entirely to the right of L within any time T>0, even when the solution is at rest at t=0 and t=T.

ABSTRACT

We consider in this paper the problem of the Lagrangian controllability for the Korteweg-de Vries equation. Using the $N$-solitons solution, we prove that, for any length of the spatial domain $L>0$ and any time $T>0$, it is possible to choose appropriate boundary controls of KdV equation such that the flow associated to this solution exit the domain in time $T$.

Motivation & Objective

• To establish small-time Lagrangian controllability for the 1D Korteweg-de Vries (KdV) equation with boundary controls.
• To demonstrate that particles initially in [0,L] can be transported entirely to the right of L within any time T>0.
• To construct a solution that is at rest at both t=0 and t=T, ensuring no residual motion after control.
• To use the asymptotic behavior of N-soliton solutions to achieve particle transport via soliton interactions and speeds.
• To ensure the solution remains small in H^2 norm near t=0 and t=T, satisfying regularity and smallness conditions for controllability.

Proposed method

• Constructs a solution of the KdV equation using N-soliton solutions in the form of traveling waves with amplitudes α_k and speeds α_k².
• Employs the change of variables x ↦ x−t, y ↦ 6η to transform the KdV equation into η_t + 6ηη_x + η_xxx = 0, simplifying soliton analysis.
• Uses the soliton solution η(x,t) = (α²/2) sech²( (−α(x−s) + α³t)/2 ) to model individual wave pulses with known speed and amplitude.
• Arranges solitons with increasing amplitudes and carefully tuned positions so that their combined flow pushes all particles in [0,L] beyond x=L.
• Applies asymptotic estimates on soliton interaction terms (e.g., f_k(ξ_k^−,t)f_{k−1}^{−1}(ξ_k^+,t)) to show that overlapping effects vanish as α_1 → ∞.
• Uses the flow equation ∂Φ/∂t = G(Φ(x,t),t) with G derived from the solution to estimate particle trajectories and ensure Φ(x,T) ≥ L.

Experimental results

Research questions

• RQ1 Can the KdV equation be Lagrangian controllable in arbitrarily small time T>0 for any domain length L>0?
• RQ2 Is it possible to drive all particles initially in [0,L] to the right of L using only boundary controls and soliton-based solutions?
• RQ3 Can the solution be constructed such that it is at rest at both t=0 and t=T, ensuring no residual motion?
• RQ4 How do soliton interactions affect the net particle transport in the KdV flow under boundary control?
• RQ5 What conditions on soliton amplitudes and positions ensure that the flow velocity remains sufficient to push particles beyond L in finite time?

Key findings

• For any T,L>0, there exists a solution y∈C([0,T];H²(0,L)) of the KdV equation with boundary controls such that the associated flow Φ satisfies Φ(x,T)≥L for all x∈[0,L].
• The solution is constructed explicitly using N-soliton solutions with increasing amplitudes α_k, where N ≈ 4Lα₁² / ln(√(2α₁)(1+√(1−1/(2α₁)))), ensuring sufficient particle transport.
• The flow velocity ∂Φ/∂t is bounded below by ∑ₖ (αₖ/16) times characteristic functions of soliton regions, ensuring net rightward movement.
• The solution satisfies ∂Φ/∂t(x,0)=0 and ∂Φ/∂t(x,T)=0 for all x∈[0,L], confirming the solution is at rest at both initial and final times.
• The H² norm of the solution is less than δ on (0,ε₁)∪(T−ε₂,T), satisfying smallness conditions near t=0 and t=T.
• As α₁→∞, the interaction terms between solitons vanish, ensuring that soliton propagation dominates and enables reliable particle transport beyond L.