Natural tangent dynamics with recurrent biorthonormalizations: a geometric computational approach to dynamical systems exhibiting slow manifolds and periodic/chaotic limit sets (Q2492248)

Consider the dynamical system \(x^{\prime }=f(x)\), \(x\in C\subset \mathbb{R}^{n}\), where \(f(x)\) is a smooth vector field belonging to the tangent space \( TC_{x}\;\)at \(x.\) If the dynamical system is characterized by a wide range of timescales (stiff problem), then the dynamics can be split into fast and slow modes. The dynamics of the slow modes can be conveniently addressed by introducing the geometric concept of slow invariant manifolds. Two important methods to approach this class of systems are: the computational singular perturbation (CSP) method and the intrinsic low-dimensional manifold (ILDM) method. However, their application is limited to the case of dynamical systems with stable equilibria, which is not always the case for dynamical systems modeling real life phenomena. The objective of the paper is to propose a geometric method, referred to as the natural tangent dynamics with recurrent biorthonormalizations (NTDRB) that unifies into a single computational technique the analysis of dynamical systems possessing either a slow manifold ( associated with an equilibrium point), or exhibiting nontrivial limit sets (limit cycles and chaotic attractors). The basic idea of the method is to follow the natural evolution of a vector and a covector basis ( defined respectively in the tangent and cotangent bundles) induced by the dynamical system itself, and to periodically apply a biorthonormalization procedure which contrasts the natural tendencies of vectors and covectors to align towards the most unstable/slow subspaces. The analysis of vector dynamics in the tangent space is a well-known tool for the computation of Lyapunov spectra, but the introduction of covector dynamics and the biorthonormalization procedure represent the original steps towards the computation of an optimal system of basis vectors and covectors, describing the invariant geometric structure of the long-term dynamics.