Published 29 Jan 2020 in math.NA, cs.NA, cs.SY, eess.SY, math.OA, and math.OC | (2001.11133v1)
Abstract: In this document, some structured operator approximation theoretical methods for system identification of nearly eventually periodic systems, are presented. Let C<sup>n×</sup>m denote the algebra of n×m complex matrices. Given $\varepsilon>0$, an arbitrary discrete-time dynamical system (Σ,T) with state-space Σ contained in the finite dimensional Hilbert space C<sup>n, whose state-transition map T:Σ×([0,∞)∩Z)→Σ is unknown or partially known, and needs to be determined based on some sampled data in a finite set Σ^=xt<em>1≤t≤m⊂Σ according to the rule T(xt,1)=x</em>t+1 for each 1≤t≤m−1, and given x∈Σ^. We study the solvability of the existence problems for two triples (p,A,φ) and (p,Aη,Φ) determined by a polynomial p∈C[z] with deg(p)≤m, a matrix root A∈C<sup>m×</sup>m and an approximate matrix root Aη∈C<sup>r×</sup>r of p(z)=0 with r≤m, two completely positive linear multiplicative maps φ:C<sup>m×</sup>m→C<sup>n×</sup>n and Φ:C<sup>r×</sup>r→C<sup>n×</sup>n, such that ∣T(x,t)−φ(A<sup>t)x∣≤ε and ∣Φ(Aη<sup>t)x−φ(A<sup>t)x∣≤ε, for each integer t≥1 such that ∣T(x,t)−y∣≤ε for some y∈Σ^. Some numerical implementations of these techniques for the reduced-order predictive simulation of dynamical systems in continuum and quantum mechanics, are outlined.