Topology from time series (Q2367671)

The aim of the paper is to give a method to extract topological properties of an invariant set (which is a topological manifold) of a dynamical system from delay-embedded data. At first, a short presentation of the topic of the simplicial homology is given. In order to build a simplicial complex associated to the experimental dynamical system, the time series are embedded in \(\mathbb{R}^ n\) and cut up into a class of overlapping patches. Then, a simplicial decomposition is deduced and an algorithm for the diagonalization of the matrix representation of the \(n\)-boundary operator is given. The diagonal entries of this matrix are then used to compute Betti numbers and Euler characteristic as topological invariants extracted from the time series. Finally these methods are applied to two experimental series.