Local convergence of Newton-like methods for degenerate eigenvalues of nonlinear eigenproblems. I. Classical algorithms (Q2514250): Difference between revisions
From MaRDI portal
Latest revision as of 14:56, 9 July 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Local convergence of Newton-like methods for degenerate eigenvalues of nonlinear eigenproblems. I. Classical algorithms |
scientific article |
Statements
Local convergence of Newton-like methods for degenerate eigenvalues of nonlinear eigenproblems. I. Classical algorithms (English)
0 references
3 February 2015
0 references
The local convergence of several classical single-vector Newton-like methods for the solution of a degenerate eigenvalue and a corresponding eigenvector of nonlinear algebraic eigenproblems of the form \(T(\lambda)v = 0\) is studied. The convergence is generally quadratic; in addition, Rayleigh functional iterative (RFI) and the Jacobi-Davidson (JD) method with appropriate use of the two-sided Rayleigh functional can achieve cubic convergence for problems with symmetry. The convergence results for simple eigenvalues also hold in the semi-simple case. The typical linear convergence of inverse iteration, standard RFI/JD and their two-side variants in the defective case is shown. For Part II see [ibid. 129, No. 2, 383--403 (2015; Zbl 1309.65060)].
0 references
single-vector Newton-like methods
0 references
degenerate eigenvalue
0 references
eigenvector of nonlinear algebraic eigenproblems
0 references
linear convergence
0 references
quadratic convergence
0 references
cubic convergence
0 references
Rayleigh functional iterative
0 references
Jacobi-Davidson method
0 references
0 references
0 references