On multisymplecticity of Sinc-Gauss-Legendre collocation discretizations for some Hamiltonian PDEs (Q924385)

This paper is concerned with the numerical solution of multisymplectic Hamiltonian partial differential equations (PDEs) that can be written in the form \[ M \partial_t {\mathbf{z}} + K \partial_x {\mathbf{z}} = \nabla_z S({\mathbf{z}}), \tag{1} \] where \( {\mathbf{z}} = {\mathbf{z}} (x, t) \in \mathbb R^d\) is the vector state variable defined for \( x \in\mathbb R\), \( t \geq 0\), \(M\) and \(K\) are \( d \times d \) skew symmetric matrices and \( S \) is a smooth scalar function. Since the Hamiltonian system (1) possess three local conservation laws: the multisymplectic, the local energy and the local momentum conservation laws, the author proposes finite difference approximations that preserve the above conservation laws. First of all, given a uniform spatial grid \( x_j = j h\), \(j=0, \pm 1, \ldots \) the spatial derivative \( \partial_x {\mathbf{z}} (x_j, t)\) at \( x= x_j\) is approximated by a linear interpolation of Sinc functions at the grid points and similarly for the right hand side of (1). Next, for time advancing, Runge-Kutta-Gauss collocation methods are proposed. A complete study of the preservation of discrete conservation laws is carried out. The proposed approach is applied to the nonlinear Dirac equation. Finally the results of some numerical experiments for Dirac's equation with a particular potential that possess travelling solitary waves are presented for the implicit midpoint rule as time advancing scheme.