Geometric integration on symmetric spaces (Q6120388): Difference between revisions

This interesting paper deals with geometric integration on symmetric space. Symmetric spaces originate from the seminal work on non-Euclidean geometries by Gauss, Bolyai, Lobachevsky, Riemann, Klein and Lie in the 19'th century. Examples are: \begin{itemize} \item The \(n\) sphere and the surface of the unit ball in \(\mathbb R^{n+1}\). \item Hyperbolic spaces with constant negative curvature. \item Symmetric positive definite matrices \(A,B\) with \(A.B=AB^{-1}A.\) \item Lie groups \(G\) with a symmetric product \(g.h=gh^{-1}g\), \(h\in H\) subsets of \(G\) closed under the operation . \item Homeogeneous space \(G/H\) where the Lie algebra of \(G\) splits as \(g=h\oplus m\) such that \(h\) is in the Lie algebra of \(H\) and \(m\) is a Lie triple system. \item The real line \(\mathbb R\) with \(x.y=2x-y\), \(x,y\in \mathbb R\). \item Gramman manifolds: The \(p\)-dimensional subspaces of \(\mathbb R^n\). \end{itemize} These spaces appear also in numerous algorithms for example for splines and various kinds of matrix computations. The paper under review studies various numerical algorithms for integrating some dynamical systems evolving on symmetric spaces. The prescribed integrators are constructed from several canonical type operations on the given symmetric space, its Lie triple system and the exponential from the Lie triple system to the symmetric space. The authors illustrate their algorithm with concrete examples. In particular for the \(n\)-sphere and the \(n\)-dimensional hyperbolic space their algorithms cost \(O(n)\) operations per step. The paper is well written with a good set of references.