Lie-Poisson methods for isospectral flows (Q2194776)

Lie-Poisson systems and isospectral flows are two well-known classes of dynamical systems. Lie-Poisson systems appear as Poisson reductions of Hamiltonian systems. The isospectral flows appear as Lax formulations of integrable systems. In this paper, spectral preserving numerical methods of arbitrary order for isospectral flows are developed. They preserve in the case of Hamiltonian isospectral flows the Lie-Poisson structure. The methods are simply constructed and avoid the use of constraints or exponential maps as for other numerical methods for isospectral flows. The authors show through the framework of Poisson reduction, that the methods are directly related to classical symplectic Runge-Kutta methods and partitioned symplectic Runge-Kutta methods. Therefore they are named as Isospectral Symplectic Runge-Kutta (IsoSyRK) methods. The methods are applied to Hamiltonian systems such as the generalized rigid body, the (periodic) Toda lattice, the Euler equations on a sphere, point vortices on a sphere, the Heisenberg spin chain, and the Bloch-Iserles flow. They are also applied to non-Hamiltonian systems as for example the Toeplitz inverse eigenvalue problem, Chu's flow on symmetric real matrices, and the Brockett flow. Through numerical examples, the efficiency and high-order accuracy of the methods is demonstrated. The MATLAB codes are available at \url{https://bitbucket.org/Milo_Viviani/iso-runge-kutta}.