Abstract:The polynomial eigenvalue problem arises in many applications and has received a great deal of attention over the last decade. The use of root-finding methods to solve the polynomial eigenvalue problem dates back to the work of Kublanovskaya (1969, 1970) and has received a resurgence due to the work of Bini and Noferini (2013). In this paper, we present a method which uses Laguerre iteration for computing the eigenvalues of a matrix polynomial. An effective method based on the numerical range is presented for computing initial estimates to the eigenvalues of a matrix polynomial. A detailed explanation of the stopping criteria is given, and it is shown that under suitable conditions we can guarantee the backward stability of the eigenvalues computed by our method. Then, robust methods are provided for computing both the right and left eigenvectors and the condition number of each eigenpair. Applications for Hessenberg and tridiagonal matrix polynomials are given and we show that both structures benefit from substantial computational savings. Finally, we present several numerical experiments to verify the accuracy of our method and its competitiveness for solving the roots of a polynomial and the tridiagonal eigenvalue problem.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is about the eigenvalue problem of matrix polynomials (polynomial eigenvalue problem, PEP). Specifically, the author proposes a method based on Laguerre iteration to calculate the eigenvalues of matrix polynomials and provides an initial estimation method based on the numerical range. In addition, the paper also discusses how to ensure the backward stability of the calculation results and provides robust methods for calculating the right and left eigenvectors. ### Eigenvalue Problem of Matrix Polynomials The eigenvalue problem of matrix polynomials can be expressed as: \[ P(\lambda) = \sum_{i = 0}^{d} \lambda^i A_i, \] where \( A_i \in \mathbb{C}^{n\times n} \) and \( A_d\neq 0 \). The eigenvalue \( \lambda \in \mathbb{C} \) satisfies \( \det(P(\lambda)) = 0 \), and the corresponding non - zero vector \( x \in \ker(P(\lambda)) \) is the eigenvector. ### Laguerre Iteration Method Laguerre iteration is an algorithm for solving polynomial roots, which has global convergence and local cubic convergence. For a given eigenvalue approximation \( \lambda \), the Laguerre iteration updates the approximation through the following formula: \[ \hat{\lambda}=\lambda - \frac{N_1}{S_1\pm\sqrt{(N_1 - 1)(N_1S_2 - S_1^2)}}, \] where \( S_1(\lambda)=\frac{p'(\lambda)}{p(\lambda)} \) and \( S_2(\lambda)=-\left(\frac{p'(\lambda)}{p(\lambda)}\right)' \), \( p(\lambda)=\det(P(\lambda)) \). ### Initial Estimation Method To improve the solution efficiency, the paper proposes a method based on the numerical range to calculate the initial estimate of eigenvalues. The numerical range is defined as: \[ W(P)=\{ \lambda \in \mathbb{C}:x^*P(\lambda)x = 0, \text{ for some non - zero vector } x \in \mathbb{C}^n\}. \] ### Backward Stability The paper proves that under appropriate conditions, the proposed Laguerre iteration method can ensure the backward stability of eigenvalue calculations. In addition, it also discusses how to handle the cases of zero eigenvalues and infinite eigenvalues. ### Applications and Experiments The paper shows the application of this method in Hessenberg and tridiagonal matrix polynomials and verifies the accuracy and competitiveness of this method through numerical experiments. In particular, this method performs well in solving the roots of high - degree polynomials and tridiagonal eigenvalue problems. In summary, the main contribution of this paper is to propose an efficient and stable method based on Laguerre iteration to solve the eigenvalue problem of matrix polynomials and verify its effectiveness and superiority through numerical experiments.

On the application of Laguerre's method to the polynomial eigenvalue problem

Numerical methods for eigenvalues of singular polynomial eigenvalue problems

A THICK-RESTART LANCZOS ALGORITHM WITH POLYNOMIAL FILTERING FOR HERMITIAN EIGENVALUE PROBLEMS

Improved Laguerre Spectral Methods with Less Round-off Errors and Better Stability

Fast and backward stable computation of the eigenvalues and eigenvectors of matrix polynomials

A Ulm-like Method for Inverse Eigenvalue Problems

Computing the Action of the Generating Function of Bernoulli Polynomials on a Matrix with An Application to Non-local Boundary Value Problems

Fast computation of spectral densities for generalized eigenvalue problems

To solving the eigenvalue problem for polynomial matrices of general form

Automatic rational approximation and linearization of nonlinear eigenvalue problems

An integral method for solving nonlinear eigenvalue problems

Randomized methods for computing joint eigenvalues, with applications to multiparameter eigenvalue problems and root finding

The infinite Lanczos method for symmetric nonlinear eigenvalue problems

Subspace embedding with random Khatri-Rao products and its application to eigensolvers

Spectral Method with the Tensor-Product Nodal Basis for the Steklov Eigenvalue Problem

Restarted Generalized Krylov Subspace Methods for Solving Large-Scale Polynomial Eigenvalue Problems

Successive linear Newton interpolation methods for solving the large-scale nonlinear eigenvalue problems

Advances in Solution of Classical Generalized Eigenvalue Problem

On matrix balancing and eigenvector computation

Inflationary dynamics for matrix eigenvalue problems

Inverse Sturm-Liouville problem with polynomials in the boundary condition and multiple eigenvalues