Multipeakons of a two-component modified Camassa-Holm equation and the relation with the finite Kac-van Moerbeke lattice (Q2629099)

The subject of this article is the two-component modified Camassa-Holm equation \[ \begin{aligned} m_t + ((u-u_x)(v+v_x)m)_x & = 0, \\ n_t + ((u-u_x)(v+v_x)n)_x & = 0, \\ m = u - u_{xx}, \quad n = v - v_{xx}.\end{aligned} \] It is known that this system allows a zero curvature representation of which the spatial component is given by the first order system \[ \Psi_x = \frac{1}{2} \begin{pmatrix} -1 & \lambda m \\ -\lambda n & 1 \end{pmatrix} \Psi. \] The authors consider a special class of multi-peakon solutions of the form \[ \begin{aligned} u(x,t) & = \sum_{k=1}^K m_{2k-1}(t) \mathrm{e}^{-|x-x_{2k-1}(t)|}, \\ v(x,t) & = \sum_{k=1}^K n_{2k}(t) \mathrm{e}^{-|x-x_{2k}(t)|}, \end{aligned} \] where \(x_1< x_2 < \ldots<x_{2K}\) and the weights \(m_{2k-1}\) and \(n_{2k}\) are positive. For this class, they first discuss the associated isospectral problem and subsequently solve the corresponding inverse problem using Stieltjes' work on continued fractions. This allows them to write down explicit formulas for these kinds of multi-peakon solutions. Furthermore, sufficient conditions for global existence are given and the long-time asymptotics are derived in this case. The final section provides a correspondence between the present peakon system and the finite Kac-van Moerbeke lattice.