Index. for generalized eigenvalue problem, butterfly form, 211

Similar documents
Numerical Methods in Matrix Computations

Matrix Algorithms. Volume II: Eigensystems. G. W. Stewart H1HJ1L. University of Maryland College Park, Maryland

Applied Linear Algebra

APPLIED NUMERICAL LINEAR ALGEBRA

Math 102, Winter Final Exam Review. Chapter 1. Matrices and Gaussian Elimination

homogeneous 71 hyperplane 10 hyperplane 34 hyperplane 69 identity map 171 identity map 186 identity map 206 identity matrix 110 identity matrix 45

Contents. Preface for the Instructor. Preface for the Student. xvii. Acknowledgments. 1 Vector Spaces 1 1.A R n and C n 2

Preliminary/Qualifying Exam in Numerical Analysis (Math 502a) Spring 2012

Structured Krylov Subspace Methods for Eigenproblems with Spectral Symmetries

Matrix Mathematics. Theory, Facts, and Formulas with Application to Linear Systems Theory. Dennis S. Bernstein

Index. Copyright (c)2007 The Society for Industrial and Applied Mathematics From: Matrix Methods in Data Mining and Pattern Recgonition By: Lars Elden

Eigenvalue Problems. Eigenvalue problems occur in many areas of science and engineering, such as structural analysis

Scientific Computing: An Introductory Survey

Eigenvalues and eigenvectors

Eigenvalue and Eigenvector Problems

AMS526: Numerical Analysis I (Numerical Linear Algebra)

Numerical Methods I Eigenvalue Problems

Preface to the Second Edition. Preface to the First Edition

EIGENVALUE PROBLEMS. EIGENVALUE PROBLEMS p. 1/4

1 Number Systems and Errors 1

Review problems for MA 54, Fall 2004.

Last Time. Social Network Graphs Betweenness. Graph Laplacian. Girvan-Newman Algorithm. Spectral Bisection

Index. Banach space 630 Basic Jordan block 378, 420

Index. book 2009/5/27 page 121. (Page numbers set in bold type indicate the definition of an entry.)

I. Multiple Choice Questions (Answer any eight)

LARGE SPARSE EIGENVALUE PROBLEMS. General Tools for Solving Large Eigen-Problems

Arnoldi Methods in SLEPc

Solution of eigenvalue problems. Subspace iteration, The symmetric Lanczos algorithm. Harmonic Ritz values, Jacobi-Davidson s method

LARGE SPARSE EIGENVALUE PROBLEMS

MATRIX AND LINEAR ALGEBR A Aided with MATLAB

QR-decomposition. The QR-decomposition of an n k matrix A, k n, is an n n unitary matrix Q and an n k upper triangular matrix R for which A = QR

Numerical Methods - Numerical Linear Algebra

Introduction to Applied Linear Algebra with MATLAB

ITERATIVE METHODS FOR SPARSE LINEAR SYSTEMS

FINITE-DIMENSIONAL LINEAR ALGEBRA

Applied Linear Algebra in Geoscience Using MATLAB

Servicepart. References 379. Index 381

Chapter 3 Transformations

Orthogonal iteration to QR

Preface to Second Edition... vii. Preface to First Edition...

Diagonalizing Matrices

EECS 275 Matrix Computation

EIGENVALUE PROBLEMS (EVP)

MATH 425-Spring 2010 HOMEWORK ASSIGNMENTS

6.4 Krylov Subspaces and Conjugate Gradients

LINEAR ALGEBRA 1, 2012-I PARTIAL EXAM 3 SOLUTIONS TO PRACTICE PROBLEMS

Math 504 (Fall 2011) 1. (*) Consider the matrices

Conceptual Questions for Review

1 9/5 Matrices, vectors, and their applications

The Eigenvalue Problem: Perturbation Theory

Solution of eigenvalue problems. Subspace iteration, The symmetric Lanczos algorithm. Harmonic Ritz values, Jacobi-Davidson s method

G1110 & 852G1 Numerical Linear Algebra

Fast Hessenberg QR Iteration for Companion Matrices

Iterative Methods for Sparse Linear Systems

Linear Algebra. Brigitte Bidégaray-Fesquet. MSIAM, September Univ. Grenoble Alpes, Laboratoire Jean Kuntzmann, Grenoble.

Ir O D = D = ( ) Section 2.6 Example 1. (Bottom of page 119) dim(v ) = dim(l(v, W )) = dim(v ) dim(f ) = dim(v )

Contents. Preface... xi. Introduction...

SUMMARY OF MATH 1600

MATH 1553, Intro to Linear Algebra FINAL EXAM STUDY GUIDE

AMS526: Numerical Analysis I (Numerical Linear Algebra)

MATH 31 - ADDITIONAL PRACTICE PROBLEMS FOR FINAL

Math 411 Preliminaries

Hands-on Matrix Algebra Using R

MATH 240 Spring, Chapter 1: Linear Equations and Matrices

AMS526: Numerical Analysis I (Numerical Linear Algebra) Lecture 23: GMRES and Other Krylov Subspace Methods; Preconditioning

Lecture Notes for Inf-Mat 3350/4350, Tom Lyche

4.8 Arnoldi Iteration, Krylov Subspaces and GMRES

Numerical Methods for Solving Large Scale Eigenvalue Problems

Scalar, Vector, and Matrix Mathematics

The value of a problem is not so much coming up with the answer as in the ideas and attempted ideas it forces on the would be solver I.N.

Introduction to Numerical Analysis

LinGloss. A glossary of linear algebra

The Lanczos and conjugate gradient algorithms

Computational Methods. Eigenvalues and Singular Values

Linear Algebra: Matrix Eigenvalue Problems

ANSWERS (5 points) Let A be a 2 2 matrix such that A =. Compute A. 2

MATH 1120 (LINEAR ALGEBRA 1), FINAL EXAM FALL 2011 SOLUTIONS TO PRACTICE VERSION

Eigenvalues and Eigenvectors

The QR Algorithm. Marco Latini. February 26, 2004

Report Number 09/32. A Hamiltonian Krylov-Schur-type method based on the symplectic Lanczos process. Peter Benner, Heike Faßbender, Martin Stoll

Matrices, Moments and Quadrature, cont d

Computing Eigenvalues and/or Eigenvectors;Part 2, The Power method and QR-algorithm

Numerical Mathematics

AMS526: Numerical Analysis I (Numerical Linear Algebra)

Computation of eigenvalues and singular values Recall that your solutions to these questions will not be collected or evaluated.

13-2 Text: 28-30; AB: 1.3.3, 3.2.3, 3.4.2, 3.5, 3.6.2; GvL Eigen2

Charles University Faculty of Mathematics and Physics DOCTORAL THESIS. Krylov subspace approximations in linear algebraic problems

Main matrix factorizations

1. General Vector Spaces

Bindel, Fall 2016 Matrix Computations (CS 6210) Notes for

ABSTRACT ALGEBRA WITH APPLICATIONS

Nonlinear palindromic eigenvalue problems and their numerical solution

Math 302 Outcome Statements Winter 2013

Krylov Subspaces. Lab 1. The Arnoldi Iteration

Parallel Singular Value Decomposition. Jiaxing Tan

Linear Algebra Primer

MATHEMATICS COMPREHENSIVE EXAM: IN-CLASS COMPONENT

MAT 610: Numerical Linear Algebra. James V. Lambers

forms Christopher Engström November 14, 2014 MAA704: Matrix factorization and canonical forms Matrix properties Matrix factorization Canonical forms

p q m p q p q m p q p Σ q 0 I p Σ 0 0, n 2p q

Transcription:

Index ad hoc shifts, 165 aggressive early deflation, 205 207 algebraic multiplicity, 35 algebraic Riccati equation, 100 Arnoldi process, 372 block, 418 Hamiltonian skew symmetric, 420 implicitly restarted, 364, 374 on a product, 412 skew Hamiltonian, 401 skew symmetric, 381 unitary (= isometric), 382 ARPACK, 364 augmented pencil, 268 backward error, 92 backward stability, 92 and residuals, 107 109 and unitary matrices, 117 balancing, 200 Bartels-Stewart algorithm, 196, 199 variant, 200 basis, 20 orthonormal, 20 standard, 20 Bauer-Fike theorem, 94 95 variant, 110 biorthonormal vectors, 390 block Krylov process, 418 blocking for efficient cache use, 132 breakdown of unsymmetric Lanczos process, 391, 397 bulge pencil, 269 272 bulge-chasing algorithm, generic, 175 177 bi-directional, 287 close-up view, 188 191 for generalized eigenvalue problem, 244 246 butterfly form, 211 cache memory, 132 efficient use in QR algorithm, 203 Cauchy-Schwarz-Bunyakovski (CSB) inequality, 4 Cayley transform, 377, 381 of a Hamiltonian matrix, 400 characteristic equation, 34 of generalized eigenvalue problem, 236 Chebyshev polynomials, 354 359 Cholesky decomposition, 16 Cholesky LR algorithm, 167 column compression, 57 column space, 23 companion matrix, 39, 142, 147 complex orthogonal matrix, 121 complex symmetric matrix, 121 condition number of eigenvalue, 97 of eigenvector, 105 of invariant subspace, 100, 105 of matrix, 6 related to SVD, 30 conjugate transpose, 3 continuity of eigenvalues, 92 Lipschitz, 96 contraction mapping theorem application of, 102 proof of, 112 convergence of GR algorithms, 223 228 of QR algorithm, 225 of Krylov process, 368 439

440 Index of subspace iteration, 215 CS decomposition, 75 cyclic matrix, 295 perfect shuffle of, 299 D-orthogonal matrix, 13 decomposition Cholesky, 16 CS,75 GR, 129 133 HR, 14, 130 Jordan, see Jordan canonical form LR, 10, 130 QR, 11, 130, 134 RG, 149 Schur, 44 singular value, 28, 305 309, 318 spectral, 73 SR, 13, 130 symplectic URV, 324, 325 defective matrix, 37, 53 deflating pair of subspaces, 258 deflation, 164 aggressive early, 205 207 of infinite eigenvalue, 254 degree of a GR iteration, 159 of a GZ iteration, 244 derogatory matrix, 183 diagonal matrix, 9 diagonalizable matrix, 37 dimension, 20 direct rotator, 89 direct sum, 21 direct vs. iterative methods, 153 distance between subspaces, 81 divide-and-conquer method, 208 dominant eigenvalue/vector, 154 dominant invariant subspace, 154 duality and Krylov subspaces, 147 in subspace iteration, 156 eigenspace, 33 eigenvalue, 33 algebraic multiplicity of, 35 associated with invariant subspace, 38 condition number of, 97 continuity of, 92 geometric multiplicity of, 69 infinite, 235, 252 257 eigenvectors, 33 left, right, 34 linear independence of, 37 elementary matrix, 118 elementary reflector, 119 elimination matrix, 117 124 equivalence of matrix pairs, strict, 236 exact shift, 183, 193 in implicit restart, 365, 367 exceptional shifts, 165 factorization, see decomposition flip matrix, 147 flop, 131 full rank, 23 Galois theory, 153, 240 Gauss transform, 118 119 with pivoting, 119 generalized eigenvalue problem, 235 as a product eigenvalue problem, 319 geometric multiplicity, 69 Gerschgorin disk theorem, 93 Givens transformation, 120 GR algorithm, 159, 175 convergence of, 223 228 cubic, 227 quadratic, 226 explicit, 159 for a product, 295 297, 301 302 implicit, 175 GR decomposition, 129 condensed, 133 operation count for, 130 Gram-Schmidt process, 134 and orthoprojectors, 138 applied twice, 135

Index 441 GZ algorithm, 243, 320 implicit, 244 reverse, 279 Hamiltonian Lanczos process, 404 as a product Krylov process, 416 stability test, 408 Hamiltonian matrix, 17, 227 eigenvalue symmetry of, 41 Schur form, 326 SR algorithm for, 209 symplectic URV decomposition of, 324, 325 HAPACK, 210 Hermitian matrix, 16, 50, 207 208, 317, 374 Hessenberg matrix, upper, 140 and GR algorithms, 162 and Krylov process, 360 and Krylov subspaces, 141 m-hessenberg matrix, 175 proper (= unreduced), 141, 175 skew Hamiltonian, 211 Hessenberg-triangular form, 239 Householder transformation, 119 HR algorithm, 209 as HZ algorithm, 249 251 cubic convergence of, 227 on a product, 339 HR decomposition, 14, 130 hyperbolic matrix, 13 elimination matrices, 123 124 HZ algorithm, 249 251 implicit restart, 352, 364 370 explained as subspace iteration, 366 367 first method, 364 second method (thick restart), 367 unitary, 387 infinite eigenvalue, 235 removal of, 254 inner product, 4 invariant subspace, 38 eigenvalues associated with, 38 existence of, 43, 44 isometric Arnoldi process, 382 isotropic subspace, 326, 400 iterative vs. direct methods, 153 Jacobi s method, 208 Jordan block, 61 Jordan canonical form, 53 65 complex, 63 practical determination, 63 64 real, 64 Kronecker canonical form, 236 Kronecker product, 65 Krylov matrix, 177 left, 281 Krylov process, 359 360 as a partial similarity transformation, 361 block, 418 convergence of, 368 implicit restart of, 364 370 on a product, 412 residual theorem, 362 Krylov sequence, 35, 352 Krylov subspaces and Hessenberg form, 141 approximating eigenvectors from, 352 defined, 141 duality and, 147 of a skew-hamiltonian matrix, 401 Krylov-Schur algorithm, 367, 374 Lanczos process, 374, 390 breakdown of, 391, 397 Hamiltonian, 404 Hamiltonian skew symmetric, 420 skew Hamiltonian, 402 skew symmetric, 381 symmetric, 374 symplectic, 408 unsymmetric, 390 LAPACK efficient cache use by, 133 QR routines, 200 207 QZ routines, 248

442 Index sep estimator, 104 linear independence, 20 Lipschitz continuity, 96 locking, 368 LR algorithm, 160, 179 and quotient-difference algorithm, 169 Cholesky variant, 167 LR decomposition, 10, 130 with partial pivoting, 10, 130 matrix companion, 39, 142, 147 complex orthogonal, 121 complex symmetric, 121 cyclic, 295 defective, 37, 53 derogatory, 183 diagonal, 9 diagonalizable, 37 elementary, 118 flip, 147 Hamiltonian, 17, 41, 209, 227, 321 342, 404 408 Hamiltonian skew symmetric, 408, 419 Hermitian, 16, 50, 207 208, 317, 374 Hessenberg, 140, 162, 360 hyperbolic, 13 Krylov, 177 left, 281 m-hessenberg, 175 nilpotent, 56 nonderogatory, 183 normal, 16, 227 orthogonal, 11 Pascal, 348 permutation, 10 positive definite, 16 pseudosymmetric, 145, 208, 227, 250, 338 quasitriangular, 45 semisimple, 37 signature, 13 skew Hamiltonian, 17, 142, 144, 211 skew Hermitian, 16 skew symmetric, 16, 317 sparse, 351 symmetric, 16, 50, 207 208, 317, 374 symplectic, 12, 41, 211, 408 411 totally nonnegative, 348 triangular, 9, 36 tridiagonal, 141 unit triangular, 9 unitary, 10, 343, 377, 382 memory hierarchy, 132 minor, leading principal, 14 MRRR method, 208 multiplicity of eigenvalue algebraic, 35 geometric, 69 nilpotent matrix, 56 nonderogatory matrix, 183 norm matrix, 5 spectral, 5 and SVD, 30 vector, 3 normal matrix, 16, 227 null space, 22 nullity, 22 oblique projector, 27 orthogonal complement, 21 orthogonal matrix, 11 complex, 121 orthonormal set, 20 orthoprojector, 27 parallel QR algorithm, 201 Pascal matrix, 348 pencil (= matrix pair), 236 augmented, 268 regular vs. singular, 236 perfect shift, 183, 193 perfect shuffle, 12, 299 permutation matrix, 10 pipelined QR iterations, 201 plane rotator, 120 positive definite matrix, 16 power method, 154

Index 443 principal angles, 76 79 relationship to SVD, 79 principal submatrix, leading, 10 product eigenvalue problem, 293 Krylov subspace methods for, 412 removal of zero eigenvalues, 304 305 role of cyclic matrix, 295 projection theorem, 20 projector, 26 oblique vs. orthogonal, 27 spectral, 72 proper upper Hessenberg matrix, 141 pseudosymmetric matrix, 145, 227, 250 and product eigenvalue problem, 338 HR algorithm for, 208 Pythagorean theorem, 5 QR algorithm, 160, 179 convergence of, 225 cubic convergence of, 227 efficient cache use in, 203 explicit, 160 for Hamiltonian matrices, 210 for the SVD, 308 for unitary matrices, 343 implicit, 179 parallel, 201 QR decomposition, 11, 130 and Gram-Schmidt process, 134 quasitriangular matrix, 45 quotient-difference (qd) algorithm, 167 171 as a product LR algorithm, 311 differential form (dqds), 170, 311 QZ algorithm, 244, 248 killed by underflow, 276 range of matrix, 22 rank, 22 full, 23 Rayleigh quotient shift, generalized, 165 real Schur theorem, 45 reorthogonalization, 373 residuals, and backward stability, 107 109 resolution of the identity, 73 RG algorithm, 278 RG decomposition, 149 Riccati equation, 100 Ritz value, 374, 380 Ritz vector, 380 rotator, 120 Schur decomposition, 44 Schur parameters, 343 Schur theorem, 43 for a Hamiltonian matrix, 326 for a skew-hamiltonian matrix, 144 generalized, 236 real, 45 semisimple matrix, 37 diagonalizability of, 53 separation (sep), 102 shift blurring, 201, 267, 273 shift-and-invert strategy, 354 shifts of origin, 155, 162 166 ad hoc, 165 convergence acceleration, 162 164 exact, 183, 193 exceptional, 165 generalized Rayleigh quotient, 165 in a GZ iteration, 243 transmission of, 267 Wilkinson, 166, 201, 208 SHIRA, 402 shuffle, perfect, 12, 299 signature matrix, 13 similarity of matrices, 42 simple eigenvalue, 96 simultaneous iteration, 156 singular value decomposition (SVD), 28 as a product eigenvalue problem, 305 309 of a product, 318 QR algorithm for, 308 related to eigenvalue problem, 52 singular vector, 29 skew-hamiltonian matrix, 17, 211 Arnoldi process for, 401 Hessenberg form of, 142 Jordan form of, 144

444 Index Lanczos process for, 402 real Schur form of, 144 skew-hermitian matrix, 16 skew-symmetric matrix, 16, 317 span, 19 spanning set, 20 sparse matrix, 351 spectral decomposition, 73 spectral projector, 72 spectral theorem for normal matrices, 48 for real symmetric matrices, 50 spectrum, 33 SR algorithm, 209 cubic convergence of, 227 for symplectic matrix, 211 reduced to HR algorithm, 341 342 SR decomposition, 13, 130 stability, see backward stability standard basis, 20 strict equivalence of matrix pairs, 236 structure preservation principle, 160 subspace, 19 invariant, 38 eigenvalues associated with, 38 isotropic, 326 Krylov, see Krylov subspaces matrix representation of, 23 subspace iteration, 153 convergence of, 215 duality in, 156 multiple steps of, 155 superlinear convergence of, 217 SVD, see singular value decomposition Sylvester equation, 54 generalized, 262, 264 use in block swapping, 195 symmetric Lanczos process, 374 block, 419 computing singular values by, 414 symmetric matrix, 16, 50, 207 208, 317, 374 complex, 121 symplectic Lanczos process, 408 stability test, 411 symplectic matrix, 12 butterfly form, 211 eigenvalue symmetry of, 41 elimination matrices, 121 123 SR algorithm for, 211 symplectic shear, 122 symplectic URV decomposition, 324, 325 tensor product, 65 thick restart, 367 totally nonnegative matrix, 348 trace, 46, 50 transpose, 3 triangle inequality, 3 triangular matrix, 9 eigenvalues of, 36 unit, 9 tridiagonal matrix, 141 ultimate shift strategy, 193 unitary matrix, 10 Arnoldi process for, 382 Cayley transform of, 377, 381 QR algorithm for, 343 unsymmetric Lanczos process, 390 breakdown of, 391, 397 real case, 393 stability test, 397 Van Loan s VZalgorithm, 320 Weyr characteristic, 61 Wilkinson shift, 166, 201, 208 Wintner-Murnaghan theorem, 45