Computational Mathematics

Prediction of multidimensional time series by method of inverse spectral problem

A. I. Sedov

South Ural State University, Chelyabinsk, Russia

Abstract: The paper develops a new method for predicting time series by the inverse spectral problem. We show that it is possible to construct a differential operator such that its eigenvalues coincide with a given numerical sequence. The paper gives a theoretical justification of the proposed method. The algorithm for finding a solution and an example of constructing a differential operator with partial derivatives are given. In this paper, we present a generalization in the case of multidimensional time series.

Keywords: Laplace operator, inverse spectral problem, eigenvalues, time series.

UDC: 517.984

Received: 25.01.2022

Language: English

DOI: 10.14529/jcem220104