pith. sign in

arxiv: math-ph/0210063 · v1 · pith:RG2HASLNnew · submitted 2002-10-31 · 🧮 math-ph · cs.NA· math.MP· math.NA

Exploring defective eigenvalue problems with the method of lifting

classification 🧮 math-ph cs.NAmath.MPmath.NA
keywords defectiveeigenvaluezeroaccuratecomputationmatrixmultiplicitynullvectors
0
0 comments X
read the original abstract

Consider an N x N matrix A for which zero is a defective eigenvalue. In this case, the algebraic multiplicity of the zero eigenvalue is greater than the geometric multiplicity. We show how an inflated (N+1) x (N+1) matrix L can be constructed as a rank one perturbation to A, such that L is singular but no longer defective, and the nullvectors of L can be easily related to the nullvectors of A. The motivation for this construction comes from linear stability analysis of an experimental reaction-diffusion system which exhibits the Turing instability. The utility of this scheme is accurate numerical computation of nullvector(s) corresponding to a defective zero eigenvalue. We show that numerical computations on L yield more accurate eigenvectors than direct computation on A.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.