# D e example 7114 t h 87 lanczos tridiagonalization

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: hat is described in Example 7.9.4 (p. 606), and describe f (A). Y P 7.10.16. Limits and Group Inversion. Given a matrix Bn×n of rank r such that index (B) ≤ 1 (i.e., index (λ = 0) ≤ 1 ), the Jordan form for B 0 looks like 0 C 0 = P−1 BP, so B = P 0 C P−1 , where C 0 0 r ×r is nonsingular. This implies that B belongs to an algebraic group G with respect to matrix multiplication, and the inverse of B in G is B# = P 0 C0 1 P−1 . Naturally, B# is called the group inverse of − 0 B. The group inverse is a special case of the Drazin inverse discussed in Example 5.10.5 on p. 399, and properties of group inversion are developed in Exercises 5.10.11–5.10.13 on p. 402. Prove that if limk→∞ Ak exists, and if B = I − A, then O C lim Ak = I − BB# . k→∞ In other words, the limiting matrix can be characterized as the diﬀerence of two identity elements— I is the identity in the multiplicative group of nonsingular matrices, and BB# is the identity element in the multiplicative group containing B. Copyright c 2000 SIAM Buy online from SIAM http://www.ec-securehost.com/SIAM/ot71.html Buy from AMAZON.com 7.10 Diﬀerence Equations, Limits, and Summability http://www.amazon.com/exec/obidos/ASIN/0898714540 641 It is illegal to print, duplicate, or distribute this material Please report violations to [email protected] 7.10.17. If Mn×n is a group matrix (i.e., if index (M) ≤ 1 ), then the group inverse of M can be characterized as the unique solution M# of the equations MM# M = M, M# MM# = M# , and MM# = M# M. In fact, some authors use these equations to deﬁne M# . Use this characterization to show that if M = BC is any full-rank factorization of ∗ M, then M# = B(CB)−2 C. In particular, if M = U1 DV1 is the full-rank factorization derived from the singular value decomposition as described in (7.10.38), then ∗ ∗ M# = U1 D−1/2 (V1 U1 )−2 D−1/2 V1 D E ∗ ∗ = U1 D−1 (V1 U1 )−2 V1 ∗ ∗ = U1 (V1 U1 )−2 D−1 V1 . T H IG R Y P O C Copyright c 2000...
View Full Document

## This document was uploaded on 03/06/2014 for the course MA 5623 at City University of Hong Kong.

Ask a homework question - tutors are online