• 検索結果がありません。

(1)THE OPTIMAL PERTURBATION BOUNDS FOR THE WEIGHTED MOORE-PENROSE INVERSE∗ WEI-WEI XU†, LI-XIA CAI‡, AND WEN LI§ Abstract

N/A
N/A
Protected

Academic year: 2022

シェア "(1)THE OPTIMAL PERTURBATION BOUNDS FOR THE WEIGHTED MOORE-PENROSE INVERSE∗ WEI-WEI XU†, LI-XIA CAI‡, AND WEN LI§ Abstract"

Copied!
18
0
0

読み込み中.... (全文を見る)

全文

(1)

THE OPTIMAL PERTURBATION BOUNDS FOR THE WEIGHTED MOORE-PENROSE INVERSE

WEI-WEI XU, LI-XIA CAI, AND WEN LI§

Abstract. In this paper, we obtain optimal perturbation bounds of the weighted Moore-Penrose inverse under the weighted unitary invariant norm, the weightedQ-norm and the weightedF-norm, and thereby extend some recent results.

Key words. Weighted Moore-Penrose inverse, Weighted unitary invariant norm, Weighted Q-norm, WeightedF-norm.

AMS subject classifications. 15A09, 15A18, 15A24.

1. Introduction. LetCm×nbe the set of complexm×nmatrices andCm×nr be the subset consisting of all matrices in Cm×n of rankr. Let A∈Cm×n. We denote kAk, kAk2, kAkQ and kAkF by the unitary invariant norm, spectral norm, Q-norm andF-norm ofA, respectively. The conjugate transformation and the Moore-Penrose generalized inverse of a matrixAare denoted byA andA, respectively.

Weighted problems, such as the weighted generalized inverse problem and the weighted least squares problem, draw more and more attention, see e.g., [2, 4, 8, 12].

A generalization of the generalized inverse is the weighted Moore-Penrose inverse of an arbitrary matrix which has many applications in numerical computation, statistics, prediction theory, control systems and analysis and curve fitting, see e.g., [1, 9, 14].

There have been many numerical methods for the computation of the weighted Moore- Penrose inverse, see e.g., [6, 7, 10, 11]. It is an interesting problem to determine how the weighted Moore-Penrose inverse is transformed under perturbation. Answers to this problem will have application in numerical computation, prediction theory and curve fitting. Therefore, it is of significance to estimate the optimal perturbation bounds of the weighted Moore-Penrose inverse. The weighted unitary invariant norm

Received by the editors on October 17, 2010. Accepted for publication on May 14, 2011. Han- dling Editor: Miroslav Fiedler.

Institute of Computational Mathematics and Scientific/Engineering Computing, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, PO Box 2719, Beijing 100190, P.R. China ([email protected]).

School of Mathematical Sciences, South China Normal University, Guangzhou, 510631 China ([email protected]).

§School of Mathematical Sciences, South China Normal University, Guangzhou, 510631 China ([email protected]). The work was supported in part by Research Fund for the Doctoral Program of Higher Education of China (Grant No. 20104407110001).

521

(2)

is a more general norm and in terms of this norm, the bounds for the weighted Moore- Penrose inverse can be characterized by weighted singular values ((M, N) singular values). Recently, much effort has been made for estimating perturbation bounds of the Moore-Penrose inverse, see e.g., [5, 9, 13, 14]. In [13], Wedin presented the perturbation bounds of the Moore-Penrose inverse under a general unitarily invariant norm, the spectral norm and Frobenius norm, respectively. Meng and Zheng in [5]

obtained the optimal perturbation bounds for the Moore-Penrose inverse under the Frobenius norm. Cai et al. in [3] obtained the additive and multiplicative perturbation bounds for the Moore-Penrose inverse under the unitary invariant norm and theQ- norm, which improves the corresponding results in [13]. In this paper, we will focus our attention on optimal perturbation bounds for the weighted Moore-Penrose inverse in the weighted unitary invariant norm, the weighted Q-norm and the weighted F- norm and thereby extend the corresponding results in [3] and [5].

We first introduce some basic definitions:

Definition 1.1. [3] A unitary invariant norm k · kis called a Q-norm if there exists another unitarily invariant norm k · k such that kYk = (kYYk)12, which is denoted byk · kQ.

Note thatF-norm and 2-norm areQ-norms.

Definition 1.2. [15] For an arbitrary matrix A ∈ Cm×n, there is a unique matrixX∈Cn×msatisfying the following equalities:

• AXA=A;

• XAX=X;

• (M AX) =M AX;

• (N XA)=N XA.

Then matrix X is called a weighted Moore-Penrose inverse of A and denoted by X =AM N. HereM andN are the given Hermitian positive definite matrices, which are called weighted matrices.

Definition 1.3. [15] LetA∈Cm×n. Then the following norms

• kAk(M N)=kM12AN12k;

• kAkF(M N)=kM12AN12kF;

• kAkQ(M N)=kM12AN12kQ;

• kAkM N =kM12AN12k2,

are called the weighted unitary invariant norm, the weightedF-norm, the weighted Q-norm and the weighted spectral norm ofA, respectively.

Definition 1.4. [15] Let A ∈ Cm×nr . The (M, N) weighted singular value de-

(3)

composition (MN-SVD) ofA∈Cm×nr is expressed as follows:

A=U

Σ 0 0 0

V=U1ΣV1, (1.1)

where U = (U1, U2) ∈ Cm×n and V = (V1, V2) ∈ Cn×n satisfy UM U = Im and VN−1V = In, Σ =diag(σ1, . . . , σr), σi = √

λi and λ1 ≥ · · · ≥ λr > 0 are the nonzero eigenvalues ofN−1AM A. Thenσ1, . . . , σr>0 are called the nonzero (M, N) weighted singular values ofA.

The rest of this paper is organized as follows. In Section 2, we give some lemmas, which are useful to deduce our main results. In Sections 3 and 4, we consider the additive and multiplicative perturbation of the weighted Moore-Penrose inverse. Some new bounds for additive and multiplicative perturbation under the normsk · k(M N), k · kQ(M N)andk · kF(M N)are presented, which extends the corresponding ones in [5]

and [13]. In Section 5, we give some numerical examples to illustrate the optimality of our given bounds under the weightedQ-norm andF-norm, respectively. Finally, in Section 6 we give concluding remarks.

2. Preliminaries. In this section we give some lemmas, which are useful to deduce our main results.

Lemma 2.1. [12] Let Ahave MN-SVD (1.1). Then (1)AM N =N−1V1Σ−1U1M;

(2)kAM NkN M = σ1

r.

Lemma 2.2. [3]Let B have the block form B=

B11 B12 B21 B22

.

Then

kBk2Q≤ kB11k2Q+kB12k2Q+kB21k2Q+kB22k2Q.

Lemma 2.3. [3]LetB1andB2be two Hermitian matrices and letP be a complex matrix. Suppose that there are two disjoint intervals separated by a gap of width at least η, where one interval contains the spectrum of B1 and the other contains that of B2. If η > 0, then there exists a unique solution X to the matrix equation B1X−XB2=P and moreover,

kXk ≤ 1 ηkPk.

(4)

Lemma 2.4. [3]Let W ∈Cn×n be a unitary matrix with the block form W =

W11 W12

W21 W22

,

where W11 ∈Cr×r,W22 ∈ C(n−r)×(n−r), 1 ≤r < n. ThenkW12k=kW21k for any unitarily invariant norm.

3. Additive perturbation bounds. In this section, we will present optimal ad- ditive perturbation bounds of the weighted Moore-Penrose inverse under the weighted unitarily invariant norm, the weightedQ-norm and the weightedF-norm, respectively.

Theorem 3.1. Let A∈Cm×nr andB=A+E∈Cm×ns . Then

kBM N −AM Nk(N M)≤(kAM NkN MkBM N kN M+ max{kAM Nk2N M, kBM Nk2N M})kEk(M N).

(3.1)

Proof. LetAandB have the following (M, N) weighted singular value decompo- sitions:

A=U

Σ1 0 0 0

V=U1Σ1V1, B= ˜U

Σ˜1 0 0 0

= ˜U1Σ˜11, (3.2)

where U = (U1, U2), ˜U = ( ˜U1,U˜2) ∈ Cm×m, V = (V1, V2), ˜V = ( ˜V1, ˜V2) ∈ Cn×n satisfy UM U = Im, U˜MU˜ = Im, VN−1V = In and ˜VN−1V˜ = In, Σ1=diag(σ1, . . . , σr), Σ˜1 =diag(˜σ1, . . . ,σ˜s) with σ1 ≥ · · · ≥σr>0 and ˜σ1≥ · · · ≥

˜ σs>0.

By (3.2) we have

E=B−A= ˜U1Σ˜11−U1Σ1V1. (3.3)

By the MN-SVDs (3.2) ofA andB we know that M12U, M12U , N˜ 12V andN12V˜ are unitary matrices. Hence from (3.3) one may deduce that

Σ˜11N−1V1−U˜1M U1Σ1= ˜U1M EN−1V1, (3.4)

U2MU˜1Σ˜1=U2M EN−11, (3.5)

Σ1V1N−12=−U1M EN−12. (3.6)

It follows from (3.4) that

1N−1V1Σ−11 −Σ˜−111M U1= ˜Σ−111M EN−1V1Σ−11 . (3.7)

(5)

By Lemma 2.1 we obtainAM N =N−1V1Σ−11 U1M andBM N =N−11Σ˜−111M.

Then, by Definition 1.3, we have

kBM N −AM Nk(N M)=kN12(N−11Σ˜−111M −N−1V1Σ−11 U1M)M12k

=kV˜N12(N121Σ˜−111M12 −N12V1Σ−11 U1M12)M12Uk

=kV˜(N−11Σ˜−111M−N−1V1Σ−11 U1M)Uk

=

12

(N−11Σ˜−111M −N−1V1Σ−11 U1M)(U1, U2)

=

Σ˜−111M U1−V˜1N−1V1Σ−11 Σ˜−111M U2

−V˜2N−1V1Σ−11 0

, (3.8)

from which one may deduce that kBM N −AM Nk(N M)

Σ˜−111M U1−V˜1N−1V1Σ−11 0

0 0

+

0 Σ˜−111M U2

−V˜2N−1V1Σ−11 0

. (3.9)

By (3.7) we have

Σ˜−111M U1−V˜1N−1V1Σ−11 0

0 0

≤ 1

σrσ˜skM12EN12k. (3.10)

By (3.5) and (3.6) we have

Σ˜−111M U2= ˜Σ−21 (U2M EN−11), V˜2N−1V1Σ−11 = (U1M EN−12)Σ−21 . Thus

0 Σ˜−111M U2

−V˜2N−1V1Σ−11 0

(3.11)

≤max{ 1 σ2r, 1

˜ σ2s}

0 (U2M EN−11)

−(U1M EN−12) 0

. Notice that

0 (U2M EN−11) (U1M EN−12) 0

12

N−1EM(U1, U2)

=V˜N−1EM U . (3.12)

(6)

SinceM12U andN12V˜ are unitary matrices, it follows from (3.11) and (3.12) that

0 Σ˜−111M U2

−V˜2N−1V1Σ−11 0

≤max{ 1 σ2r, 1

˜

σ2s}N12EM12

= max{ 1 σ2r, 1

˜ σ2s}

M12EN12

= max{ 1 σ2r, 1

˜

σ2s} kEk(M N), which together with (3.9), (3.10) and Lemma 2.1(2) deduces (3.1).

Remark 3.2. If we take M =N=I, then Theorem 3.1 reduces to kB−Ak ≤(kAk2kBk2+ max{kAk22,kBk22})kEk, (3.13)

which is the result of Theorem 3.1 in [3].

For theQ(N M)-norm we provide the following bound.

Theorem 3.3. Let A∈Cm×nr andB=A+E∈Cm×ns . Then kBM N−AM NkQ(N M)

(3.14)

≤ q

kAM Nk4N M +kBM N k4N M+kAM Nk2N MkBM N k2N MkEkQ(M N).

Proof. The bound (3.14) follows immediately from Lemma 2.2, (3.4)–(3.6) and (3.8).

Remark 3.4. If we take M =N=I, then Theorem 3.3 reduces to kB−AkQ

q

kAk42+kBk42+kAk22kBk22kEkQ, which is the result of Theorem 3.2 in [3].

Theorem 3.5. Let A∈Cm×nr andB=A+E∈Cm×ns . Then

kBM N −AM NkF(N M)≤max{kAM Nk2N M,kBM Nk2N M}kEkF(M N). (3.15)

Proof. It follows from (3.3) that

U1MU˜1Σ˜1−Σ1V1N−11=U1M EN−11, (3.16)

2M U1Σ1=−U˜2M EN−1V1, U2MU˜1Σ˜1=U2M EN−11, (3.17)

Σ˜11N−1V2= ˜U1M EN−1V2, Σ1V1N−12=−U1M EN−12. (3.18)

(7)

By (3.16), we have

Σ−11 U1MU˜1−V1N−11Σ˜−11 = Σ−11 U1M EN−11Σ˜−11 . (3.19)

It is easy to see that

kBM N −AM NkF(N M)=kN12(N−11Σ˜−111M−N−1V1Σ−11 U1M)M12kF

=kVN12(N121Σ˜−111M12 −N12V1Σ−11 U1M12)M12U˜kF

=kV(N−11Σ˜−111M−N−1V1Σ−11 U1M) ˜UkF

=

V1 V2

(N−11Σ˜−111M−N−1V1Σ−11 U1M)( ˜U1,U˜2)

F

=

V1N−11Σ˜−11 −Σ−11 U1MU˜1 −Σ−11 U1MU˜2

V2N−11Σ˜−11 0

F

(3.20) .

It follows from (3.8) and (3.20) that

2kBM N −AM Nk2F(N M)=kΣ˜−111M U1−V˜1N−1V1Σ−11 k2F +kΣ˜−111M U2k2F

+kV˜2N−1V1Σ−11 k2F+kV1N−11Σ˜−11 −Σ−11 U1MU˜1k2F

+kΣ−11 U1MU˜2k2F+kV2N−11Σ˜−11 k2F, which together with Lemma 2.2, (3.4)–(3.6) and (3.16)–(3.19) yields

2kBM N −AM Nk2F(N M)=kΣ˜−111M EN−1V1Σ−11 k2F+kΣ˜−211N−1EM U2k2F

+kΣ−11 U1M EN−11Σ˜−11 k2F+kV˜2N−1EM U1Σ−21 k2F

+kΣ−21 V1N−1EMU˜2k2F +kV2N−1EMU˜1Σ˜−21 k2F

(3.21)

≤ 1

σ2rσ˜s2(kU˜1M EN−1V1k2F+kU1M EN−11k2F) + 1

σ4r(kV˜2N−1EM U1k2F+kV1N−1EMU˜2k2F)

≤max{ 1 σ4r, 1

˜

σ4s}(kU˜1M EN−1V1k2F+kU1M EN−11k2F) +kV˜1N−1EM U2k2F+kV2N−1EMU˜1k2F

+kV˜2N−1EM U1k2F+kV1N−1EMU˜2k2F

≤2 max{ 1 σ4r, 1

˜

σ4s}kM12EN12k2F

= 2 max{kAM Nk4N M,kBM N k4N M}kEk2F(M N). Therefore,

kBM N −AM NkF(N M)≤max{kAM Nk2N M,kBM Nk2N M}kEkF(M N), which implies (3.15) holds.

(8)

Remark 3.6. If we take M =N=I, then Theorem 3.5 reduces to kB−AkF ≤max{kAk22,kBk22}kEkF,

which is the result of Theorem 2.1 in [5].

Now we consider the case that rank(A) =rank(B), i.e.,A, B∈Cm×nr . Theorem 3.7. Let A, B=A+E∈Cm×nr . Then

kBM N −AM Nk(N M)≤[kAM NkN MkBM N kN M+ (kAM NkN M+kBM NkN M) min{kAM NkN M,kBM N kN M}]kEk(M N).

Proof. Let

D1=

Σ1 0 0 σrI

, D2=

Σ˜1 0 0 σ˜rI

, and

X =

Σ˜−111M U1−V˜1N−1V1Σ−11 Σ˜−111M U2

−V˜2N−1V1Σ−11 0

.

Then

XD1+D2X =

Σ˜−111M U1Σ1−V˜1N−1V1 σrΣ˜−111M U2

−V˜2N−1V1 0

+

1M U1−Σ˜11N−1V1Σ−111M U2

−σ˜r2N−1V1Σ−11 0 (3.22) .

From (3.4) it is easy to see that

kΣ˜−111M U1Σ1−V˜1N−1V1k ≤ 1

˜

σrkM12EN12k ≤ 1

˜

σrkEk(M N), (3.23)

kU˜1M U1−Σ˜11N−1V1Σ−11 k ≤ 1

σrkM12EN12k ≤ 1

σrkEk(M N). (3.24)

SinceUeM U is unitary, by Lemma 2.4 we have kUe1M U2k=kUe2M U1k. By (3.3) and (3.5), one may deduce that

Ue2M U1=−Ue2M EN−1V1Σ−11

(9)

and

Ue1M U2=Σe−11 (U2M EN−1Ve1). This in turn implies that

kUe2M U1k ≤ 1

σrkEkM N, and

kUe1M U2k ≤ 1 f

σrkEkM N, respectively. Then

kU˜1M U2k ≤ 1

max{σr,σ˜r}kEk(M N), and thus

rΣ˜−111M U2k ≤ σr

˜

σrkU˜1M U2k ≤ σr

˜

σrmax{σr,σ˜r}kEk(M N). By an analogous argument, we have

kV˜2N−1V1k ≤ 1

max{σr,σ˜r}kEk(M N)

and

kσ˜r2N−1V1Σ−11 k ≤ σ˜r

σrkV˜2N−1V1k ≤ σ˜r

σrmax{σr,˜σr}kEk(M N), which together with (3.8), (3.22)–(3.24) and Lemma 2.3 give the desired result.

Remark 3.8. If we take M =N=I, then Theorem 3.7 reduces to kB−Ak ≤[kAk2kBk2+ (kAk2+kBk2) min{kAk2,kBk2}]kEk, (3.25)

which is the result of Theorem 3.3 in [3].

For the weightedF-norm we have:

Theorem 3.9. Let A, B=A+E∈Cm×nr . Then

kBM N−AM NkF(N M)≤ kAM NkN MkBM N kN MkEkF(M N). (3.26)

Proof. Since in (1.1)UMU ,˜ V N˜ −1V andVN−1V˜ are unitary, by Lemma 2.4 we have

kU˜1M U2kF =kU˜2M U1kF,kU2MU˜1kF =kU1MU˜2kF, (3.27)

(10)

kV˜1N−1V2kF =kV˜2N−1V1kF,kV2N−11kF =kV1N−12kF. (3.28)

It follows from (3.17), (3.18) and (3.27), (3.28) that

2kBM N −AM Nk2F(N M)=kΣ˜−111M EN−1V1Σ−11 k2F+kΣ−11 U1M EN−11Σ˜−11 k2F

+ 1

˜

σr2(kU˜1M U2k2F+kV2N−11k2F) + 1

σr2(kV˜2N−1V1k2F+kU1MU˜2k2F)

≤ 1

˜

σr2σ2r(kU˜1M EN−1V1k2F +kU1M EN−11k2F) + 1

˜

σr2(kU˜2M U1k2F+kV1N−12k2F) + 1

σr2(kV˜1N−1V2k2F+kU2MU˜1k2F)

= 1

˜

σr2σ2r(kU˜1M EN−1V1k2F +kU1M EN−11k2F) + 1

˜

σr2(kU˜2M EN−1V1Σ−11 k2F+kΣ−11 U1M EN−12k2F) + 1

σr2(kΣ˜−111M EN−1V2k2F+kU2M EN−11Σ˜−11 k2F)

≤ 2

˜

σr2σ2rkM12EN12k2F = 2kAM Nk2N MkBM Nk2N MkEk2F(M N), which implies (3.26) holds.

Remark 3.10. Since

kAM NkN MkBM N kN M ≤max{kAM Nk2N M,kBM N k2N M}, the bound is sharper than the one in (3.15).

For the weightedQ-norm we have:

Theorem 3.11. LetA, B =A+E∈Cm×nr . Then kBM N−AM NkQ(N M)≤√

3kAM NkN MkBM N kN MkEkQ(M N). (3.29)

Proof. By (3.20) we have

kBM N −AM NkQ(N M)=

V1N−11Σ˜−11 −Σ−11 U1MU˜1 −Σ−11 U1MU˜2

V2N−11Σ˜−11 0

Q.

(11)

It follows from (3.8), (3.27), (3.28) and Lemma 2.2 that

2kBM N −AM Nk2Q(N M)≤ kΣ˜−111M U1−V˜1N−1V1Σ−11 k2Q+kΣ˜−111M U2k2Q

+kV˜2N−1V1Σ−11 k2Q+kV1N−11Σ˜−11 −Σ−11 U1MU˜1k2Q

+kΣ−11 U1MU˜2k2Q+kV2N−11Σ˜−11 k2Q, which together with (3.4)–(3.7) and (3.16)–(3.19) gives that

2kBM N −AM Nk2Q(N M)≤ kΣ˜−111M EN−1V1Σ−11 k2Q+kΣ−11 U1M EN−11Σ˜−11 k2Q

+ 1

˜

σ2r(kU˜1M U2k2Q+kV2N−11k2Q) + 1

σ2r(kV˜2N−1V1k2Q+kU1MU˜2k2Q)

≤ 1

σ2rσ˜r2(kU˜1M EN−1V1k2Q+kU1M EN−11k2Q) + 1

˜

σ2r(kU˜2M U1k2Q+kV1N−12k2Q) + 1

σ2r(kV˜1N−1V2k2Q+kU2MU˜1k2Q)

= 1

σ2rσ˜r2(kU˜1M EN−1V1k2Q+kU1M EN−11k2Q) + 1

˜

σ2r(kU˜2M EN−1V1Σ−11 k2Q+kΣ−11 U1M EN−12k2Q) + 1

σ2r(kΣ˜−111M EN−1V2k2Q+kU2M EN−11Σ˜−11 k2Q)

≤ 1

σ2rσ˜r2(kU˜1M EN−1V1k2Q+kU˜2M EN−1V1k2Q

+kU˜1M EN−1V2k2Q+kU1M EN−11k2Q

+kU1M EN−12k2Q+kU2M EN−11k2Q)

≤6kAM Nk2N MkBM Nk2N MkEk2Q(M N), which implies that the inequality (3.29) holds.

Remark 3.12. If we take M =N =I, then Theorem 3.11 reduces to kB−AkQ≤√

3kAk2kBk2kEkQ, which is the result of Theorem 3.4 in [3].

It is easy to see that

√3kAM NkN MkBM N kN M ≤ q

kAM Nk4N M+kBM N k4N M+kAM Nk2N MkBM Nk2N M, which implies that the bound in (3.29) is sharper than the one in (3.14).

(12)

4. Multiplicative perturbation bounds. In this section, we present optimal multiplicative perturbation bounds of the weighted Moore-Penrose inverse. Let B be a multiplicative perturbed matrix ofA, i.e., B =D1AD2, where D1 and D2 are m×mand n×nnonsingular matrices, then rank(A) =rank(B).

Theorem 4.1. Let A∈Cm×n andB =D1AD2, where D1 andD2 are respec- tively m×m andn×nnonsingular matrices. Then

kBM N −AM Nk(N M)≤max{kAM NkN M,kBM N kN M1(D1, D2), (4.1)

where

Ψ1(D1, D2) =kIm−D1k(M M)+kIm−D−∗1 k(M M)+kIn−D2k(N N)+kIn−D2−1k(N N).

Proof. LetA andB have the MN-SVDs (3.2). Clearly we have B−A=B(In−D2−1) + (D1−Im)A

= (Im−D1−∗)B+A(D2−In).

(4.2)

It follows from (3.3) and (4.2) that

1Σ˜11−U1Σ1V1= ˜U1Σ˜11(In−D2−1) + (D1−Im)U1Σ1V1

= (Im−D1−∗) ˜U1Σ˜11+U1Σ1V1(D2−In).

(4.3)

By (4.3) we obtain

Σ˜11N−1V1−U˜1M U1Σ1= ˜Σ11(In−D2−1)N−1V1+ ˜U1M(D1−Im)U1Σ1,

U1MU˜1Σ˜1−Σ1V1N−11=U1M(Im−D1−∗) ˜U1Σ˜1+ Σ1V1(D2−In)N−11,

U2MU˜1Σ˜1=U2M(Im−D−∗1 ) ˜U1Σ˜1, U˜2M U1Σ1=−U˜2M(D1−Im)U1Σ1, Σ˜11N−1V2= ˜Σ11(In−D2−1)N−1V2,

Σ1V1N−12=−Σ1V1(D2−In)N−12, from which one may deduce that

1N−1V1Σ−11 −Σ˜−111M U1

(4.4)

(13)

= ˜V1(In−D−12 )N−1V1Σ−11 + ˜Σ−111M(D1−Im)U1, Σ˜−111M U2= ˜Σ−111(Im−D−11 )M U2,

(4.5)

−V˜2N−1V1Σ−11 = ˜V2N−1(D2−In)V1Σ−11 , (4.6)

Σ−11 U1MU˜1−V1N−11Σ˜−11 (4.7)

= Σ−11 U1M(Im−D1−∗) ˜U1+V1(D2−In)N−11Σ˜−11 ,

−Σ−11 U1MU˜2= Σ−11 U1(D1−Im)MU˜2, (4.8)

V2N−11Σ˜−11 =V2N−1(In−D2−∗) ˜V1Σ˜−11 . (4.9)

By (3.7) and (4.4)–(4.6) we obtain kBM N −AM Nk(N M)

=

L Σ˜−111(Im−D1−1)M U2

2N−1(D2−In)V1Σ−11 0

−Σ˜−111M(D1−Im)U1 Σ˜−111(Im−D−11 )M U2

0 0

+

−V˜1(In−D−12 )N−1V1Σ−11 0 V˜2N−1(D2−In)V1Σ−11 0

M Σ˜−111M12M12(Im−D1−1)M12M12U2

0 0

! +

−V˜1N12N12(In−D2−1)N12N12V1Σ−11 0 V˜2N12N12(D2−In)N12N12V1Σ−11 0

!, (4.10)

where

L ≡ −V˜1(In−D−12 )N−1V1Σ−11 −Σ˜−111M(D1−Im)U1, M ≡ −Σ˜−111M12M12(D1−Im)M12M12U1. This implies (4.1) holds.

Remark 4.2. If we take M =N=I, then Theorem 4.1 reduces to kB−Ak ≤max{kAk2,kBk2}Ψ(D1, D2),

参照

関連したドキュメント