Reduced-order equations are simplified models that capture the essential features of some process originally described by a more complicated set of governing equations. In this study, we highlight the effectiveness of a nonlinear least-squares algorithm in deriving one-dimensional tight-binding equations of topological insulator models from noisy spectral data. We show that applying fits to bands with uncorrelated disorder yields convergence to the underlying clean model at the probabilistic rate O(N−1/2), where N is the number of realizations. Moreover, for not too large a disorder, the fitted models reproduce the appropriate topology and presence of localized edge states. This approach has the ability to conveniently derive reduced-order models from realistic data.
Nameika et al. (Wed,) studied this question.