An Iterative Eigenvector Technique for Optimization Analysis,
Abstract
Certain structural optimization analyses require the repeated generation of sparse, symmetric linear systems of equations for which a few eigenpairs (eigenvalues and corresponding eigenvectors) are to be determined. Typically, these linear systems have similar spectra. Consequently, eigenpairs for one system tend to be reasonably close approximations to those of another next linear system. An iterative eigenvector analysis technique that utilizes the available approximate eigenpairs in order to reduce computation costs is described. it is based on Rayleigh quotient iteration with a Lanczos type iterative equation solver. A rational transformation of the initial forms of the linear systems is used in order to overcome the adverse matrix conditions that typically ruin the convergence of the iterative equation solver. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1981
- Accession Number
- ADP000035
Entities
People
- Paul S. Jensen
Organizations
- Lockheed Martin Missiles and Space