BLOCK TOEPLITZ MATRICES: SOME BASIC

RESULTS
arXiv:2103.14827v1 [math.FA] 27 Mar 2021

MUHAMMAD AHSAN KHAN

Abstract. Given A, B, C and D block Toeplitz matrices, we will

prove some of the basic results concerning the product AB − CD.
In addition, with respect to change of basis, the characterization
of normal block Toeplitz matrices with entries in the algebra of
diagonal matrices is also obtained.

1. Introduction
A scalar Toeplitz matrix is an n × n matrix with the following struc-
ture:
a0 a1 a2 . . . an−1
 
 a−1 a0 a1 . . . an−2 
a a a0 . . . an−3  .
 
A =  −2
 −1
 ... .. .. .. 
. . . 
a1−n a2−n a3−n . . . a0
The entries depend upon the difference i−j and hence they are constant
down all the diagonals. These matrices are ubiquitous and play an
important role in different fields of mathematics, as well as applied
areas as signal processing or time series analysis. The subject is many
decades old; among monograph dedicated to the subject are [12, 16],
and [10].
The related area of block Toeplitz matrices is less studied, one of the
reason being the new difficulties that appear with respect to the scalar
case. Besides its theoretical interests, the subject is also important to
multivariate control theory. As references for block Toeplitz matrices
on can use [1, 11, 13, 14], and [15]. In [2] the authors have proved the
variety of algebraic results about Toeplitz matrices. They have formu-
lated those algebraic results in terms of products of Toeplitz matrices.
Given Toeplitz matrices A, B, C, and D, their main result determines
if the matrix AB − CD is Toeplitz. The necessary and sufficient con-
dition is a rank two matrix equation involving tensor products of the
Key words and phrases. Toeplitz matrix, Displacement matrix, Block Toeplitz
matrix .
1
2 MUHAMMAD AHSAN KHAN

vectors defining A, B, C, and D. They have also proved the necessary

and sufficient condition for AB − CD = 0.
The authors in [2] have also completely characterized the normal Toeplitz
matrices. The characterization of normal Toeplitz matrices has been
discussed in [3, 4, 6, 7, 18].
The problem of characterizing normal block Toeplitz matrices when
their entries are arbitrary scalar matrices of dimension d is an open
problem and there seems to be no chances to traet it in full generality.
In this article we classified normal block Toeplitz matrices when the
entries are taken from the algebra of d × d scalar diagonal matrices.
The purpose of the present paper is to generalize some of the reults of [2]
concerning the product of block Toeplitz matrices by way of introduc-
ing the special structure of the displacement matrix of a Block Toeplitz
matrix. The second task is to obtain the necessary and sufficient con-
dition for a block Toeplitz matrix to be a normal block Toeplitz matrix
with entries from the algebra of scalar diagonal matrices.
The paper is organized as follows : Notations and some basic facts
about displacement matrices are presented in section 2. Then as we
will be interested in block Toeplitz matrices, some basic properties
concerning the product of these matrices are derived in section 3. The
last section is devoted to the study of normal Block Toeplitz matrices
with entries from the fixed algebra of scalar diagonal matrices.

2. Preliminaries
Let C denote the set of complex numbers. We designate by Mn
the algebra of n × n matrices and by Dd the algebra of d × d diagonal
matrices with entries from C. Throughout in this paper we prefer to
n−1
label the indices from 0 to n − 1; so A ∈ Mn is written A = (ai,j )i,j=0
with ai,j ∈ C. Then Tn ⊂ Mn is the space of Toeplitz matrices A =
n−1
(ai−j )i,j=0 . We will mostly be interested in block matrices, that is,
matrices whose elements are not necessarily scalars, but elements in
Md . Thus a block Toeplitz matrix is actually an nd × nd matrix, but
which has been decomposed in n2 blocks of dimension d, and these
blocks are constant parallel to the main diagonal. we will use the
following notations:
• Mn ⊗Md is the collection of n×n block matrices whose entries
all belong to Md ;
• Tn ⊗Md is the collection of n×n block Toeplitz matrices whose
entries all belong to Md ;
• Dn ⊗ Md is the collection of n × n diagonal block Toeplitz
matrices whose entries all belong to Md .
 BLOCK TOEPLITZ MATRICES: SOME BASIC RESULTS 3

• C1 ⊗Md is the collection of all n×1 block matrices whose entries

all belong to Md ;
• R1 ⊗ Md is the collection of all 1 × n block matrices whose
entries all belong to Md .
Obviously Dn ⊗ Md ⊂ Tn ⊗ Md ⊂ Mn ⊗ Md . For block diagonal
matrices we will use the notation
A1 0 . . . 0
 

 0 A2 . . . 0 
diag A1 A2 · · · Ad =   ... .. . . . 
. . .. 
0 0 . . . Ad
In many cases it will suffice to consider Toeplitz matrices with zero
diagonals. In other cases Ã+ Ã0 will describe the most general Toeplitz
matrix , where à is a Toeplitz matrix with 0 on the main diagonal and
Ã0 is the
 diagonal Toeplitz matrix. 
0 0 0
   
 a−1  a1−n   a1 
If a =  ... , then we define â =  ...  and if b =  ... , then
    

a1−n a−1 an−1

0
 
an−1 
b̂ = 
 ... 


a1
Let I ∈ Md be the identity matrix and S ∈ Mn ⊗ Md consisting of
zeroe matrices, except for identity matrices along the subdiagonal,i.e,
0 0 0 ... 0 0
 
I 0 0 . . . 0 0
0 I 0 . . . 0 0
 
S= . . . .
 .. .. .. . . ... 


0 0 0 ... I 0
For any M ∈ Mn ⊗ Md the displacement matrix is defined as
△(M) = M − SMS ∗ .
We will use this matrix to determine whether the difference of the
matrix products is Toeplitz. See [9] and [19] for other types of dis-
 T
I
0 
placement matrices. We denote the matrix  ...  ∈ R1 ⊗ Md by P+ ,


0
then its adjoint is the matrix P+∗ ∈ C1 ⊗ Md .
4 MUHAMMAD AHSAN KHAN

n−1
X
Lemma 2.1. If M ∈ Mn ⊗ Md then M = S k (△(M))S k∗ .
k=0

Proof.
n−1
X n−1
X
k k∗
S (△(M))S = S k (M − SMS ∗ )S k∗
k=0 k=0
n−1
X
= (S k MS k∗ − S k+1 MS k+1∗ = M − S n MS n∗ = M
k=0

Thus to determine M = 0, it is sufficient to study the simpler equa-
tion △(M) = 0
.
3. Block Toepliz product
In this section we will generalize some important reults of [2]. We
first obtain the necessary and a sufficient condition for a block matrix
A to be a Toeplitz matrix.
The following Lemma describes, △(M) is particularly simple if A is
Toeplitz.
Lemma 3.1. A ∈ Mn ⊗ Md is Toeplitz if and only if there exist
X, X ′ ∈ C1 ⊗ Md such that △(A) = XP+ + P+∗ X ′∗ .
n−1
Proof. Suppose that A = (Ai−j )i,j=0 ∈ Tn ⊗Md . Since the displacement
matrix for A is defined as △(A) = A−SAS ∗ . Then simple computation
yields that
A0 A1 A2 . . . An−1
 
 A−1 0 0 ... 0 
A−2 0 0 ... 0 
 
A=  .
 .. .. .. . . 
. . . 
A−(n−1) 0 0 . . . 0
A0 0
   
 A−1  A
 and X ′ =  . 1  then one can easily check
 
If we take X =  .
 ..   .. 
A1−n An−1
that
△(A) = XP+ + P+∗ X ′∗ .

n−1
For the converse, let A = (Ai,j )i,j=0 ∈ Mn ⊗ Md . Suppose then that
 BLOCK TOEPLITZ MATRICES: SOME BASIC RESULTS 5
 ′ 
X0 X0
 
 X1   X1′ 
X =  ..  and X =  .. 
  ′  be matrices in C1 ⊗ Md , since we
. . 
′
Xn−1 Xn−1
have
△(A) = XP+ + P+∗ X ′∗
=⇒
A = SAS ∗ + XP+ + P+∗ X ∗′
=⇒
X0 + X0′∗ X1′∗ X2′∗ . . . ′∗
Xn−1
 
 X1 A0,0 A0,1 . . . A0,n−2 
A= .. .. .. .. 
 . . . . 
Xn−1 An−2,0 An−1,1 . . . An−2,n−2
Compairing corresponding entries yields Ai1 ,j1 = Ai2 ,j2 , whenever i1 −
j1 = i2 − j2 , where 0 ≤ i1 , i2 , j1 , j2 ≤ n − 1., i.e., A is a block Toeplitz
matrix.

n−1
In the rest of this section if A = (Ai−j )i,j=0 is any block Toeplitz
matrix then for simplification we write A = Ã+
Ã0 , where à ∈ Tn ⊗Md
with A0 = 0 and Ã0 = diag A0 A0 · · · A0 .
T
0 0 0
    
 A−1   B1  An−1 
Lemma 3.2. Let A− =   ... , B+ =  ...  , A+ =  ... 
    

A−(n−1) Bn−1 A1
T
0

B1−n 
and B− =  ...  . If A = Ã + Ã0 and B = B̃ + B̃0 then there exist


B−1
Y, Y ∈ C1 ⊗ Md such that △(AB) = Y P+ + P+∗ Y ′∗ + A− B+ − A+ B− .
′

Proof. We have
△(AB) = △[(Ã + Ã0 )(B̃ + B̃0 )]
= △[ÃB̃ + ÃB̃0 + Ã0 B̃ + Ã0 B̃0 ]
= △(ÃB̃) + △(ÃB̃0 ) + △(Ã0 B̃) + △(Ã0 B̃0 )
Since B̃0 ∈ Dn ⊗ Md then ÃB̃0 ∈ Tn ⊗ Md , therefore by Lemma
3.1 there exist U, U ′ ∈ C1 ⊗ Md such that △(ÃB̃0 ) = UP+ + P+∗ U ′∗ .
6 MUHAMMAD AHSAN KHAN

Similarly △(Ã0 B̃) = V P+ + P+∗ V ′∗ and △(Ã0 B̃0 ) = W P+ + P+∗ W ′∗ ,

with V, V ′ , W, W ′ ∈ C1 ⊗ Md .
(3.1) =⇒ △(AB) = △(ÃB̃)+(U +V +W )P+ +(U ′∗ +V ′∗ +W ′∗ )P+∗
Now we have entry at the position (i, j) of ÃB̃ is
n−1
X
(3.2) (ÃB̃)i,j = Ak−i Bj−k , 0 ≤ i, j ≤ n − 1,
k=0

Using the formula (3.2) and definition of S, one obtains

(3.3) 
A−i Bj − An−i Bj−n for 1 ≤ i, j ≤ n − 1,

∗
(ÃB̃ − S ÃB̃S )i,j = (ÃB̃)0,j for 1 ≤ j ≤ n − 1,

(ÃB̃)
i,0 for 0 ≤ i ≤ n − 1.
Therefore,
(3.4) △ (ÃB̃) = XP+ + P+∗ X ′∗ + A− B+ − A+ B− .
   
(ÃB̃)0,0 0
 (ÃB̃)   (ÃB̃)0,0 
1,0 
Where X =  , X = 
′
..  . Combining (3.1) and
  
..
 .   . 
(ÃB̃)n−1,0 (ÃB̃)0,n−1
(3.4) yields
△(AB) = (U + V + W + X)P+ + (U ′∗ + V ′∗ + W ′∗ + X ′∗ )P+∗ + A− B+ − A+ B−
= Y P+ + P+∗ Y ′∗ + A− B+ − A+ B− .
Where Y = U + V + W + X0 and Y ′∗ = U ′∗ + V ′∗ + W ′∗ + X0′∗ . 
The next result is the important result of this paper.
Theorem 3.3. Let Ã− , B˜− ∈ C1 ⊗ Md and A˜+ , B˜+ ∈ R1 ⊗ Md
with 0 in the zeroth component. If A = Ã + Ã0 , B = B̃ + B̃0 ,
C = C̃ + C̃0 and D = D̃ + D̃0 then AB − CD ∈ Tn ⊗ Md if and
only if A− B+ − A+ B− = C− D+ − C+ D− .

Proof. By Lemma 3.2

(3.5)
△(AB) − △(CD) = (Y − Z)P+ + P+∗ (Y ′∗ − Z ′∗ ) + A− B+ − A+ B− − C− D+ + C+ D−
Note that the last four terms on the right side of the above equation
involve vectors with 0 in the zeroth component. By Lemma 3.1, AB −
CD is Toeplitz if and only if A− B+ − A+ B− = C− D+ − C+ D− .

 BLOCK TOEPLITZ MATRICES: SOME BASIC RESULTS 7

Corollary 3.4. AB ∈ Tn ⊗ Md if and only if A− B+ − A+ B− = 0.

.

Proof. The statement follows from Theorem 3.3 (i) by taking C = D =

0. 
Theorem 3.5. Let X, X ′ , Y and Y ′ be vectors in C1 ⊗Md with 0 on the
zero component of X ′ and Y ′ . If AB −CD ∈ Tn ⊗Md , then AB = CD
if and only if X = Y and X ′ = Y ′ .
Proof. If AB −CD ∈ Tn ⊗Md , then by Lemma 3.1 there exist X, X ′ , Y
and Y ′ in C1 ⊗ Md , with 0 along the zero component of X ′ , Y ′ such
that
△(AB) − △(CD) = (Y − Z)P+ + P+∗ (Y ′∗ − Z ′∗ ).
Now AB = CD if and only if △(AB − CD) = 0. The latter equation
holds if and only if the vectors which form the product with P+ are 0.
That is, AB = CD if and only if X = Y and X ′ = Y ′ . 

4. Normal block Toeplitz matrices

In this section we will obtain the necessary and sufficient condition
for A ∈ Tn ⊗ Dd to be a normal block Toeplitz matrix .
The next result from [2] characaterize normal Toeplitz matrices among
all Toeplitz matrices.
n−1
Theorem 4.1. Let A = (ai−j )i,j=0 ∈ Tn then A is normal if and only
¯
if either a = λb̂ for some |λ| = 1 or a = λb̂ for some |λ| = 1.
Before starting the main work of this section we first need to quote
the result from [13].
n−1
Lemma 4.2. Suppose A = (Ai−j )i,j=0 ∈ Tn ⊗ Dd . Then there is a
change of basis that brings A into the following form


A′ = diag A′1 A′2 · · · A′d ,
where for every k = 1, 2, · · · d, A′k ∈ Td .
The following result is the main result about the normal block Toeplitz
matrices.
Theorem 4.3. If A ∈ Tn ⊗ Dd then A is normal if and only if there
exist scalars λ1 , λ2 , · · · λd with |λk | = 1, such that for any k = 1, 2, · · · d,
either ak = λk or ak = λk a˜k .
8 MUHAMMAD AHSAN KHAN

Proof. If A ∈ Tn ⊗ Dd then by Lemma 4.2, A has the form

A′ = diag A′1 A′2 · · · A′d ,
where for every k = 1, 2, · · · d, A′k = (ar−s,k )dr,s=0. Now AA∗ = A∗ A if
and only if A′k A′∗
k = Ak Ak . Then by Theorem 4.1 each Ak is normal
′∗ ′

if and only if there exist scalars λ1 , λ2 , · · · λd with |λk | = 1 such that

a−1,k a1,k
   
a−2,k  a 
either ak = λk b̂ or ak = λk b̂k , where ak =  , and bk =  2,k
.  ... ,

 .. 
a−d,k ad,k
k = 1, 2, · · · d.


Acknowledgements
References
[1] A. Böttcher and B. Silbermann, Analysis of Toeplitz operators, Springer-
Verlag, 1990.
[2] C. Gu, L. Patton, Commutation relation for Toeplitz and Hankel matrices,
SIAM J. Matrix Anal. Appl. 24 (2003), 728–746.
[3] D. Farenick, M. Krupnik, N. Krupnik, W. Lee, Normal Toeplitz matrices,
SIAM J. Matrix Anal. Appl., 17 (1996), 1037-1043.
[4] D. Ikramov, On a description of normal Toeplitz matrices, Comput. Math.
Math. Phys., 34 (1994), 399–404.
[5] D. Ikramov, Classification of normal Toeplitz matrices with real elements,
Math. Notes, 57 (1995), 463-469.
[6] D. Ikramov, N. Chugunov, A criterion for the normality of a complex Toeplitz
matrix, Comput. Math. Math. Phys., 36 (1996), pp. 131–137.
[7] D. Ikramov, N. Chugunov, On the skew-symmetric part of Toeplitz matrices,
Math. Notes, 63 (1998), 124–127.
[8] D.L. Zimmerman: Block Toeplitz products of block Toeplitz matrices, Linear
and Multilinear Algebra 25 (1989), 185–190.
[9] G . Heinig, K. Rost, Algebraic Methods for Toeplitz-Like Matrices and Oper-
ators, Birkhauser Verlag, Basel, 1984.
[10] H. Widom, “Toeplitz Matrices,” in Studies in Real and Complex Analysis, (J.
I.I. Hirschmann, ed.), MAA Studies in Mathematics, Prentice-Hall, Englewood
Cliffs, NJ, 1965.
[11] I. Gohberg, S. Goldberg and M.A. Kaashoek, Classes of Linear Operators, Vol.
2, Birkhauser Verlag Basel, 1993.
[12] I. S. Iohvidov, Hankel and Toeplitz Matrices and Forms, Birkhuser Boston,
Cambridge, MA, 1982.
[13] M. A. Khan, A Family of Maximal Algebras of Block Toeplitz Matrices, An.
St. Univ. Ovidius, Constanta Seria Matematica 26 (3) (2018), 127-142.
[14] M. A. Khan, D. Timotin, Algebras of block Toeplitz Matrices with commuting
entries, Linear and Multilinear Algebra DOI10.1080/03081087.2019.1693955.
 BLOCK TOEPLITZ MATRICES: SOME BASIC RESULTS 9

[15] T. Shalom, On Algebras of Toeplitz matrices, Linear Algebra and its applica-
tions 96 (1987), 211-226.
[16] U. Grenander and G. Szego, Toeplitz Forms and Their Applications. University
of Calif. Press, Berkeley and Los Angeles, 1958.
[17] V. Gel‘fgat, A normality criterion for Toeplitz matrices, Comput. Math. Math.
Phys., 35 (1995), 1147-1150.
[18] K. Ito, Every normal Toeplitz matrix is either of type I or of type II, SIAM J.
Matrix Anal. Appl., 17 (1996), 998–1006.
[19] T. Kailath, H. Sayed, Displacement structure: Theory and applications, SIAM
Rev., 37 (1995), 297–386.

Abdus Salam School of Mathematical Sciences, GC University La-

hore, Pakistan
Email address: [email protected], [email protected]