没有合适的资源?快使用搜索试试~ 我知道了~
On the iterative bisymmetric solution of general coupled matrix ...
1 下载量 195 浏览量
2020-03-10
03:52:15
上传
评论
收藏 329KB PDF 举报
温馨提示
On the iterative bisymmetric solution of general coupled matrix equations,李东平,缪树鑫,A matrix A = (ai;j) 2 Rn×n is called bisymmetric matrix if ai;j = aj;i =an+1-j;n+1-i holds for all 1≤i; j≤n. In this paper, an efficient algorithm is presented to find the bis
资源推荐
资源详情
资源评论
On the iterative
bisymmetric solution of
general coupled matrix equations
∗
Dongping Li
a†
, Shu-Xin Miao
a,b,
, Bing Zheng
a
a
School of Mathematics and Statistics, Lanzhou University,
Lanzhou 730000, P.R. China
b
College of Mathematics and Information Science, North-West Normal University,
Lanzhou 730070, P.R. China
Abstract. A matrix A = (a
i,j
) ∈ R
n×n
is called bisymmetric matrix if a
i,j
= a
j,i
=
a
n+1−j,n+1−i
holds for all 1 ≤ i, j ≤ n. In this paper, an efficient algorithm is presented
to find the bisymmetric solution of the general coupled matrix equations. When the
general coupled matrix equations is consistent on bisymmetric solutions, then for any
initial bisymmetric matrix group, a group of bisymmetric solution can be obtained
within finite iterative steps in the absence of round off errors, and the least norm
bisymmetric solution can be obtained by cho osing a group of special kind of initial
matrices. Finally, we test the algorithm and show it’s effectiveness by a numerical
example.
Keywords: The general coupled matrix equations; Least norm solution; Bisym-
metric solution.
AMS subject classifications: 15A18, 65F10
1 Introduction
Throughout this paper, we denote by R
m×n
, SR
n×n
and BSR
n×n
the set of m × n real
matrices, n × n real symmetric matrices and n × n real bisymmetric matrices, respectively.
We use A
T
to denote the transpose of matrix A. S
n
(S
n
= (e
n
, e
n−1
, . . . , e
1
)) refers to the
∗
Supp orted by the start-up fund of Lanzhou University and the Natural Science Foundation of Gansu
Province (3ZS051-A25-020), P.R. China.
†
Corresp onding author. E-mail address: lidping06@lzu.cn(D.-P. Li).
1
http://www.paper.edu.cn
n × n reverse unite matrix (e
i
denotes ith column of n × n unite matrix). In the space
R
m×n
, we define an inner product as: hA, Bi = tr ace(B
T
A) for all A, B ∈ R
m×n
. Then the
norm of matrix A generated by this inner product is, obviously, the Frobenius norm and
denoted by k A k. For two matrix M and N, M
N
N is their kronecker product. For two
m × n matrices X and Y with X = [x
1
, x
2
, . . . x
n
] ∈ R
m×n
, col[X] is a mn column vector
consisting of columns of X, i.e.
col[X] =
³
x
T
1
, x
T
2
, · · · , x
T
n
´
T
∈ R
mn×1
and
col[X, Y ] =
Ã
col[X]
col[Y ]
!
∈ R
2mn×1
.
The matrix equations play important roles in systems and control theory[14]. Var-
ious solutions such as general, symmetric, bisymmetric and skewsymmetric solutions to
a special matrix equation were studied widely, see [1–4, 7]. The basic approaches taken
are to transform the matrix equations into form for which solutions may be readily com-
puted, such as the generalized singular decompositions (GSVD) [1, 2, 5], Hessenberg-Schur
form [6], the canonical correlation decomposition (CCD) [10] of matrices and so on. But,
these methods require compute some additional matrix transformation and decomposition,
the computational efforts rapidly increase with the dimensions of matrices to be solved and
excessive computer memory is required for computation and inversion of large matrices.
In view of these reasons, some iterative methods have been proposed to solve the matrix
equations [8,9, 12].
Our interest here is to present an efficient iterative algorithm to find the bisymmetric
solution of more general coupled matrix equations
A
11
X
1
B
11
+ A
12
X
2
B
12
+ · · · + A
1n
X
n
B
1n
= C
1
,
A
21
X
1
B
21
+ A
22
X
2
B
22
+ · · · + A
2n
X
n
B
2n
= C
2
,
.
.
.
A
m1
X
1
B
m1
+ A
m2
X
2
B
m2
+ · · · + A
mn
X
n
B
mn
= C
m
.
(1.1)
with A
ij
∈ R
p
i
×n
j
, B
ij
∈ R
n
j
×q
i
, C
i
∈ R
p
i
×q
i
, i = 1, 2, · · · , m, j = 1, 2, · · · , n.
When the general coupled matrix equations (1.1) has the bisymmetric solutions, then, for
any group of initial matrices [X
1,0
, X
2,0
, . . . , X
n,0
] with X
i,0
∈ BSR
n
i
×n
i
, i = 1, 2, . . . , n, the
sequence of matrix groups {[X
1,k
, X
2,k
, . . . , X
n,k
]} generated by the algorithm 2.1 converges
to a group of bisymmetric solutions within at most min{
P
m
i=1
p
i
q
i
,
P
n
i=1
m
i
n
i
} +1 iterative
steps in the absence of round-off errors. Moreover, if we choose the initial values as
X
j,0
=
m
X
i=1
(A
ij
T
H
i
B
ij
T
+ B
ij
H
i
T
A
ij
+ S
n
j
A
T
ij
H
i
B
ij
T
S
n
j
+ S
n
j
B
ij
H
T
i
A
ij
S
n
j
),
2
http://www.paper.edu.cn
j=1,2,. . . ,n, where H
i
, i = 1, 2, . . . , m are the arbitrary matrices with compatible size,
then the solution group [X
1
∗
, X
2
∗
, . . . , X
n
∗
] obtained by the iterative method is the least
Frobenius norm bisymmetric solution group.
The left of this paper is organized as follows. In Section 2, we propose an efficient
iterative algorithm to obtain bisymmetric solution of the coupled matrix equations (1.1)
and some properties of the algorithm are studied. In Section 3, we present an example to
illustrate the effectiveness of the algorithm proposed in the paper. Finally, we offer some
concluding remarks in Section 4.
2 The iterative method
The proposed iterative method for obtaining a group of the bisymmetric solutions of the
general coupled matrix equations (1.1) is given as follows.
Algorithm 2.1
1. Input matrices A
ij
∈ R
p
i
×n
j
, B
ij
∈ R
n
j
×q
i
, C
i
∈ R
p
i
×q
i
, X
j,0
∈ BSR
n
j
×n
j
,
i = 1, 2, · · · , m, j = 1, 2, · · · , n.
2. Calculate
R
i,0
= C
i
−
P
n
j=1
A
ij
X
j,0
B
ij
, i = 1, 2, · · · , m;
¯
P
j,0
=
P
m
i=1
A
ij
T
R
i,0
B
ij
T
, j = 1, 2, · · · , n;
P
j,0
=
1
4
[
¯
P
j,0
+
¯
P
T
j,0
+ S
n
j
(
¯
P
j,0
+
¯
P
T
j,0
)S
n
j
], j = 1, 2, · · · , n;
k := 0;
3. If R
i,k
= 0, (i = 1, 2, · · · , m) then stop;
else k := k + 1;
4. Calculate
σ
k
=
P
m
i=1
kR
i,k−1
k
2
P
n
i=1
kP
i,k−1
k
2
;
X
j,k
= X
j,k−1
+ σ
k
P
j,k−1
, j = 1, 2, · · · , n;
R
i,k
= C
i
−
P
n
j=1
A
ij
X
j,k
B
ij
= R
i,k−1
− σ
k
P
n
j=1
A
ij
P
j,k−1
B
ij
, i = 1, 2, · · · , m;
δ
k
=
P
m
i=1
kR
i,k
k
2
P
m
i=1
kR
i,k−1
k
2
;
¯
P
j,k
=
P
m
i=1
A
ij
T
R
i,k
B
ij
T
, j = 1, 2, · · · , n;
P
j,k
=
1
4
[
¯
P
j,k
+
¯
P
T
j,k
+ S
n
j
(
¯
P
j,k
+
¯
P
T
j,k
)S
n
j
] + δ
k
P
j,k−1
, j = 1, 2, · · · , n;
5. Go to Step 3.
3
http://www.paper.edu.cn
剩余15页未读,继续阅读
资源评论
weixin_38621630
- 粉丝: 3
- 资源: 914
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功