de
en
Schliessen
Detailsuche
Bibliotheken
Projekt
Impressum
Datenschutz
de
en
Schliessen
Impressum
Datenschutz
zum Inhalt
Detailsuche
Schnellsuche:
OK
Ergebnisliste
Titel
Titel
Inhalt
Inhalt
Seite
Seite
Im Dokument suchen
Integration based solvers for standard and generalized Hermitian eigenvalue problems / von Lukas Krämer. 2014
Inhalt
Motivation and outline
Introduction
Basics: (Computational) Linear Algebra
Matrices and vectors
Norms
Scalar products and orthogonality
Matrix induced scalar products and norms
Projectors
Singular value decomposition
Computer arithmetic
Eigenvalues and eigenvectors
Basic notions
Eigenspaces
Angles between vectors and subspaces
Scalar products and geometry
Angles between subspaces
Angles in mat B-induced scalar products
Eigenproblems and their numerical solution
Types of eigenproblems
Types of eigensolvers
Measures for the quality of an eigensolver
Accuracy
Reliability
General theory of contour integration based eigensolvers
Subspace eigensolvers
Rayleigh–Ritz-method
Subspace iteration
Eigenvalue bounds
Convergence of Ritz vectors
Residual based bounds
Harmonic Rayleigh–Ritz
A few facts from complex analysis
Numerical integration
Basics
Interpolatory quadrature
Gauß quadrature
Error statements
Integration of periodic functions
Eigensolvers based on integration
Literature review
Spectral projectors and resolvent
Computing an eigenspace
Error analysis of integration based eigensolvers
Introduction
Error in the integration—Trapezoidal rule
Error in the integration—Gauß–Legendre
Choice of integration contour
Influence of error in linear systems
Conclusion
FEAST eigensolver
Basic algorithm
Counting eigenvalues and size of search space
Problems with wrongly chosen mest
The selection function
Convergence rate
Eigenvalues of bu
Efficient computation of a basis for the search space
Preprocessing of FEAST
Alternatives and further discussion
Numerical experiments
Numerical integration revisited
Approximation by integration methods
Integration by approximation methods
Polynomial approximation
Introduction
Chebyshev approximation
Error estimation
Error at the boundary of ilambda
Experiments with Chebyshev-FEAST
Connection of polynomial degree and convergence rate
Adaptive choice of polynomial degree
Generalized problem
Why Chebyshev? (Other polynomials)
Transforming the integration region
Use of integral transformation
Conformal transformation of integration region
Numerical experiments and discussion
Miscellaneous issues
Linear systems
Parallelism
Orthogonality
Stopping criteria and eigenpair locking
Integration error/convergence of eigenvalues and subspaces
Conclusion
Conclusion and outlook
Index
Summary of Notation
List of Figures
List of Tables
List of Algorithms
Bibliography