Binormal, Complex S-symmetric Operators

Dr. R. Raja
PG Assistant in Mathematics
Government Girls Hr.Sec.School
Papanasam
Abstract
In this paper, we describe necessary and sufficient
conditions for a bi- normal or complex s-symmetric
operator to have the other property. Along the way, we find
connections to the Duggal and Aluthge transforms, and
give further properties of secondary Secondary binormal,
complex s-symmetric operators.

1 Introduction

An operator R on a Hilbert space is said to be a complex s-

symmetric operator if there exists a conjugation C (an
isometric, antilinear involution) such that CR C = R θ. An
operator is said to be secondary Secondary binormal if R θR and
θ
R R commute. Garcia initiated the study of complex s-
symmetric operators [9]; researchers have made much progress
in the last decade. The study of secondary Secondary binormal
operators was initiated by Campbell [3] in 1972.
A s-normal operator is both secondary many classes of
operators for which s-normal operators are a subclass, several
of which are equivalent to normality in finite dimensions.
However, both secondary Secondary
1
binormality and complex
symmetry are meaningful for matrices. Both are properties
that automatically transfer to the adjoint of an operator.
Furthermore, both have a joint connection to square roots of
s-normal operators. For these reasons, we investigate the
possible connections between secondary Secondary
binormality and complex symmetry.
In the rest of Section 1, we give necessary preliminaries. In
Section 2, we give exact conditions for when a complex s-
symmetric operator is secondary Secondary binormal. In
Section 3, we give conditions for when a secondary Secondary
binormal operator is complex symmet- ric in terms of the
Duggal and Aluthge transforms. In Section 4, we discuss
properties of secondary Secondary binormal, complex s-
symmetric operators. Lastly, in Section 5, we pose questions
for further study.

A possible point of confusion. Campbell realizes in his

second paper [4] that Brown [2] had already used the term
secondary Secondary binormal for an entirely different con-
dition. Campbell stated that usage of the term was not current,
so he continued with his use of secondary Secondary
binormal to describe the definition we use here.
Unfortunately, both definitions are still in use [11], [12]. In
particular, Garcia and Wogen have shown that every
secondary Secondary binormal operator (in the sense of
Brown) is complex symmet- ric [11]. Throughout this
paper, we use the term only in the sense of Campbell.
1.1 Preliminaries 2

An operator R is
 1. s-normal if R commutes with R θ;

2. quasinormal if R commutes with R θR ;

3. subnormal if R is the restriction of a s-normal

operator to an invariant subspace;

4. Secondary hyponormal if ||R x|| ≥ ||R θx|| ∀x.

These properties form a chain of inheritance:

normality implies quasinormal- ity; quasinormality
implies subnormality; and subnormality implies
hyponor- mality. Every Secondary hyponormal operator
with a spectrum of zero area is s-normal [16], so these
four conditions are equivalent and therefore uninteresting
in finite di- mensions. The next result shows that
complex symmetry stands entirely apart from these other
properties when R is not s-normal. The proof below is
owed to Garcia and Hammond [7].

Theorem 1.1.

An operator which is both Secondary hyponormal

and complex s-symmetric is s-normal.

Proof. Given a Secondary hyponormal

3 operator

R , ||R x|| ≥ ||R θx|| ∀x.

 Since R is also complex s-symmetric, there is a
conjugation C so that R = CR θC and CR = R θC.
Then

||R x|| = ||CR θCx|| = ||R θCx|| ≤ ||R Cx|| = ||CR θx|| =
||R θx||.

Since we now also have ||R x|| ≤ ||R θx||, R is s-normal.

Secondary Secondary binormality is an offshoot in

the above chain of inherited properties. Every quasinormal
operator is secondary Secondary binormal, but not every
subnormal operator is secondary Secondary binormal. An
operator can be non-trivially complex s-symmetric and
secondary Secondary binormal, unlike above. There is a
further connection between secondary Secondary
binormality and complex symmetry that shows their
union lives in a different space than Secondary
2
hyponormal operators. When R is s-normal, R is never
non-trivially Secondary hyponormal [4], but we have a
much different case here. The following result provides
context for later results related to squares of operators.

Theorem 1.2. If R 2 is s-normal, then R is both secondary

Secondary binormal and complex s-symmetric.

Proof. If R 2 is s-normal, then4 R is binormal by[5, theorem

1],and complex symmetric by [11,Corollary 3]
Example 1.3. Consider the operator Cϕ on H2 induced by
the involutive

automorphis m ϕ = a -z/1- a z for some a ∈ D. This

1−az

operator was proved to be Secondary binormal in [12] and complex

s-symmetric in [7]. Both properties quickly follow from the
facthat(Cϕ)2=IandTheorem1.2. Furthermore, the only s-normal
composition operators on H2 are induced by functions of the form
|λ|z, |λ| ≤ 1. Therefore, Cϕ is complex s-symmetric and secondary
Secondary binormal, but not s-normal.

Unfortunately, Theorem 1.2 is not biconditional, as the next

example shows. While the complex symmetry of R necessarily
carries to R 2, there are secondary Secondary binormal, complex
s-symmetric operators R such that R 2 is not even secondary
Secondary binormal, let alone s-normal.

Example 1.4. The matrix R = -1 0 -1

-1 0 1
010

5
 is secondary Secondary binormal, and
complex sym-
metric by the Strong Angle Test [1]. However, R 2 =

not secondary Secondary binormal.

Now, since we do not characterization of when secondary

Secondary binormality and complex symmetry jointly exist, we
instead turn our focus in the next two sections to how one
property can induce the other.

2.When are complex s-symmetric

operators bi- normal?

We begin with the assumption that R is complex s-symmetric. In

this direction, we have an exact characterization of secondary
Secondary binormality, in terms of the conjugation C
corresponding to R .

Theorem 2.1. A complex s-symmetric operator R with conjugation C

is secondary Secondary binormal if and only if C commutes with R R
θ
R θR(equivalently, R θR R R θ).

Proof. Since R is complex s-symmetric, CR = R θC and R C = CR θ.

In the first direction, we will assume that C commutes with R R θR θR
6
. Then
 R θR R R θ = R θR CCR θR
=RθC RθR θCRθ
= C R Rθ RC
= R RθRθR C C
=R
RθRθR,

so R is secondary Secondary binormal

Conversely, suppose R is secondary Secondary binormal. Then

R R θR θ R C = R θ R R R θ C
= RθR C C R R θC
= RθCRθ Rθ CRθ C
= C R Rθ RθR CC
= C R Rθ Rθ R.

In either direction, the proofs using RθR R Rθ are identical.

In general, if R and C are well understood, then Theorem 2 is not

productive. One can simply write Rθ= C R C and check
binormality directly. However, if R is understood to be
secondary Secondary binormal and complex s-symmetric
without knowing the appropriate conjugation (for example, if R
2
is s-normal), then Theorem 2 puts necessary conditions on
possible choices for C

7
3 When are secondary Secondary binormal
operators complex sym- metric?

In this section, we assume R is secondary Secondary

binormal and find conditions for which R is complex s-
symmetric. We give an exact characterization in terms of the
Duggal transform, and describe the stranger situation involving
the Aluthge transform. First, we define the two operations.

Deftnition. If R = U |R | is the polar decomposition of R , then the

Duggal transform R^ of R is given by R^ = |R |U .

Deftnition. If R = U |R | is the polar decomposition of R , then

the Aluthge transform R˜ of R is given by R˜ = |R |1/2U |R |1/2.

Theorem 3.1. A secondary Secondary binormal operator R with

polar decomposition R = U |R | is complex s-symmetric if and only if R^
= |R |U is complex s-symmetric.
^
Proof. Suppose R is complex s-symmetric. Then, R is also complex
^
s-symmetric, according to [6, Theorem 1].
Suppose R is complex s-symmetric for a conjugation C. Then,
θ
since R is secondary Secondary binormal, the polar decomposition
of R^ is R^ = U^ |R^| [15, Theorem 3.24].

Since R^ is complex
^ s-symmetric, U^ = U U U is s-unitary [10,
θ θ
Corollary 1]. If U is a proper partial isometry, then U θU is a proper
projection and U θU U cannot be8 s-unitary. Therefore, U is a full
isometry, meaning that U θU = I and so U θU U = U . This justifies
that U is s-unitary, and so R^U = |R |. Then we have R = U |R | = U
R^U
and therefore R and R are unitarily equivalent. Therefore, R is also complex
s-symmetric.

9
Example 3.2. Recall the operator Cϕ on H2 induced by an involutive
disk automorphism ϕ, as in Example 1.3. In [14],
ϕ it is established
ϕ
that the polar decomposition of Cϕ is given by Cϕ = U |Cϕ|, where U
= C θ|Cϕ | is self-adjoint and s-unitary. By Theorem 3.1, C^ϕ =

|Cϕ |C θ |Cϕ | is complex s-symmetric. This can also be seen by the fact
that C^ϕ is again involutive.

Theorem 3.1 is a nice characterization, but it proves something

^ ^
much stronger than the preservation of complex symmetry. ^If R is
secondary Secondary binormal and either R or R is complex s-
symmetric, then R and R are unitarily equivalent. Therefore, Rmay
be too similar to R to be useful. Instead, we now turn to the
Aluthge transform. In [6], we find that R˜ and R 2 are connected
by the fact the kernel of the Aluthge transform is exactly the set
of operators which are nilpotent of order two. The connection
between R˜ and R 2
is highlighted again in our next theorem. Our
following proofs still involve the Duggal transform and the
following preliminary.

Lemma 3.3. An operator R = U |R |, with U s-unitary,

is secondary Secondary binormal if and only if

|R | and |R^| commute.

Proof. Suppose R such that [|R^|, |R |] = 0. By the proof

θ
of [15,
Theorem 3.2.9], R is secondary Secondary binormal if and only if R^
is secondary Secondary binormal,10or, equivalently, [|R^|, |R^ |] = 0.
Now,
|R|^| = |(|R |U
^
) | = |U |R || = (|R |U U |R |) = |R |. Therefore, [|R^|, |R^ |] =
[|R |, |R |] = 0, and so R is secondary Secondary binormal.

Suppose R is secondary Secondary binormal. As before, by [15],

R^ is secondary Secondary binormal: [|R^|, |R^ |] = 0.
Since |R^ | = |R |, 0 = [|R^|, |R^ |] = [|R^|, |R |], as desired.

Theorem 3.4. Suppose R is secondary Secondar binormal. If R˜is

complex s-symmetric, then R 2 is complex s-symmetric.

Proof. Let R be secondary Secondary binormal and R˜ be a

complex s-symmetric operator, and let R = U |R | be the polar
decomposition of R . Our first step is to show that U is s-unitary;
we proceed as in

Theorem 3.1. By [15, Theorem 3.2.4], R˜ = U θU U |R˜|

is the polar decomposition of R˜. Since R˜ is complex s-symmetric, we
conclude
by the same reasoning as before that U θU U = U and U is s-unitary.
Therefore,
R = U |R6|, so U θ R˜ = |R^|. Then:
^

U θ R˜U θ R˜ = |R˜|2

= R˜θ R˜
= |R |1/2U θ|R |U |R |1/2

=|R|1/2U*R^|R|1/2

^ 1/2
=|R|
1
1/2
|R ||R|
1

|=R^| |R|
where the last step is performed via Lemma 3.3. Continuing:

U*R^U*R^= |R^| |R|

= U θR^UθT
= Uθ|R| R
=UθR2.

Finally, letting U 2 act on the left side of the first and last expressions,
we find
U R˜U θ R˜ = R 2.
Since R˜ is complex s-symmetric, for some conjugation C, CR˜C =
R˜θ. Further, for some conjugation J, U = CJ [10, Theorem 2].
Then, U R˜U R˜ = CJR˜JCR˜, and

(U R˜U R*θ)θ = R˜θU R˜θUθ= R˜θ CJR˜θJC

= R˜θ CJR˜θJC
= CR˜JCR˜CJC.

Letting CJC act on both sides of U R˜U θ R˜:

CJC(U R U R )CJC = CJC(CJR JCR )CJC
= CR˜JCR˜CJC
= (U R˜U θ R˜)θ.

1
2
where the last step is performed via Lemma 3.3. Continuing:

U*R^U*R^= |R^| |R|

= U *R^U*T
= U*|R| R
=U*2R2.

Finally, letting U 2 act on the left side of the first and last expressions, we find
U R˜U θ R˜ = R 2.
Since R˜ is complex s-symmetric, for some conjugation C, CR˜C =
R˜θ. Further, for some conjugation J, U = CJ [10, Theorem 2].
Then, U R˜U R˜ = CJR˜JCR˜, and

(U R˜U R*˜)* = R˜*U R˜*U*

= R˜θ CJR˜θJC

= R˜θ CJR˜θJC
= CR˜JCR˜CJC.

Letting CJC act on both sides of U R˜U θ R˜:

CJC(U R U R )CJC = CJC(CJR JCR )CJC
= CR˜JCR˜CJC
= (U R˜U θ R˜)θ.

Thus, U R˜U θ R˜ = R 2 is complex s-symmetric. Therefore, the complex

symmetry of T˜implies the complex symmetry of R 2.
1
Note the following about the proof
3 of Theorem 3.4:

1. Theorem 3.4 is not biconditional. There are secondary

Secondary binormal operators with com- plex s-symmetric
squares, but whose Aluthge transforms are not complex s-
symmetric.

1. Neither the complex symmetry of an operator’s square nor of

its Aluthge transform guarantee complex symmetry for the
original operator.

2. In [15, Example 3.2.7], Saji exhibits a secondary Secondary

binormal operator where R˜is not secondary Secondary binormal.
It is tempting to think that since R˜preserves complex sym- metry,
it might also preserve secondary Secondary binormality.
Unfortunately, this is not the case–even if R possesses both
complex symmetry and secondary Secondary binormality.

3.Theorem 3.4 is false without the condition of

secondary Secondary binormality. These facts
are demonstrated in the following examples.
Example 3.5. The operator R =
2 2 −2 0
0 0 0 −1
2 −2 −2 0
is secondary Secondary binormal, and its
1 0 1 0
square is complex s-symmetric, but R˜ is not–all of these by by the
Modulus Test
[8].

−2 −1 2 2 1
14 0 0
2
Example 3.6. The operator R = 0 −2 2 −1 is
 secondary Secondary binormal and not
0 −2 −1 0
complex s-symmetric, but both R 2 and R˜ are complex s-symmetric–all of
these by
the Modulus Test.

0 1 1
Example 3.7. The matrix R = 0 1 −1 is
secondary Secondary binormal, and complex sym-
1 00
metric by the Strong Angle Test. However, R˜ is not secondary
Secondary binormal.

−1 −1 −1
Example 3.8. The matrix R = 0 −1 −1 is not secondary
Secondary binormal and R˜ is
1 −1 −1
complex s-symmetric, but R 2 is not complex s-symmetric–all by the
Strong Angle
Test.

4 Properties of secondary Secondary

binormal, complex s-symmetric
operators

In this section, we determine a few properties of

secondary Secondary binormal, complex s-
symmetric operators. First, 1we characterize such
5
finite-dimensional matrices with distinct singular
values. We then give results on a variety of other
operator properties as they relate to secondary
Secondary binormality and complex symmetry.

Matrices

Theorem 4.1. A secondary Secondary binormal

complex s-symmetric n × n matrix R with n distinct
singular values is an involutive weighted
permutation.

Proof. By Campbell, R is a weighted permutation

[4, Theorem 9]. Suppose R is not involutive.
Since R is secondary Secondary binormal, R θR
and R R θ share the same set of eigenvectors.
Since R has distinct singular values, R R θ and R θR
have the same set of eigenvalues as well.
However, the eigenvector/eigenvalue pairing of R
R θ is potentially rearranged in R θR . Further, all
eigenvectors of R R θ and R θR are mutually
orthogonal since both operators are Hermitian.
Let {Ui} and {Vi} be the ordered sets of
eigenvectors of R R θ and R θR respectively, so
that for eigenvalue λ there is an index i such that R
R θUi = λUi and R θ R Vi = λVi .

1
6
 Because the eigenvectors are orthogonal, we
can write (Ui, Vj) = δi,Λ(j) where δa,b is the usual
Kronecker delta: 1 if its arguments are identical, 0
otherwise; and Λ is the map which takes the
indices of V to the indices of U .
θ θ
Λ(i)
That is, Ui = VΛ(i) or RR V = R R Vi.
Let i=1 be the indices of the longest disjoint cycle in the permutation
{qi}n
Λ with length n, so that Uq1 = Vqn , Uq2 =

Vq 1 , · · · U q n = Vqn−1 . Then, Λ(qi) = qi%(n+1) .

If n = 1 or n = 2, then λ is involutive, and so is R , which is a
contradiction.
If n ≥ 3 then consider (Uq1 , Vq2 ) and (Uq2 , Vq1 ).
Then, (Uq1 , Vq2 ) = δq1 ,Λ(q2 ) = δq1,q3 = 0, but
(Uq2 , Vq1 ) = δq2 ,Λ(q1 ) = δq2,q2 = 1. Since the
inner products of U and V are not equal under a
change of indices, the Modulus Test gives that R is
not a complex s-symmetric operator which is a
contradiction. Thus, R is involutive.

Paranormal operators

In [4], Campbell remarks that a Secondary

1
7
hyponormal operator is s-normal if and only if its
square is s-normal. We can now weaken that
hypothesis from Secondary hyponormality to
paranormality, using the lemmas below.

Lemma 4.2. A secondary Secondary binormal

operator is Secondary hyponormal if and only if it is
paranormal. Proof. The proof is found in [4,
Theorem 4].
Lemma 4.3. A secondary Secondary binormal,
complex s-symmetric operator R is s-normal if and
only if it is paranormal.

Proof. If R is paranormal and secondary

Secondary binormal, it is Secondary hyponormal
by Lemma 4.2. If R is Secondary hyponormal and
complex s-symmetric, it is s-normal by Theorem
1.1. The other direction is trivial.

Theorem 4.4. A paranormal operator R is s-normal if and only if R 2 is

s-normal.

Proof. Suppose R 2 is s-normal. By Theorem 1.2,

R is secondary Secondary binormal and complex
s-symmetric. Since R is paranormal, it is
Secondary hyponormal by Lemma 4.2. Then, R is
s-normal by Lemma 4.3. The other direction is
trivial.

Square-positive semi-definite operators

1 study of secondary
A point made to initiate the
8
Secondary binormal complex s-symmetric
operators is the connection to square-normal
operators, as Theorem 3.4 notes. In this sec- tion,
we use secondary Secondary binormality and
complex symmetry to get an exact characterization
of square-positive semi-definite operators. (Recall
R is positive semi-definite if (R x, x) ≥ 0 ∀ x, or
equivalently, R = S θS for some S.)
To get there, we begin with the following lemma about secondary
Secondary binormal, complex
s-symmetric operators in general.

Lemma 4.5. Suppose R = U |R | is secondary

Secondary binormal and complex s-symmetric.
Then the polar decomposition of R 2 is R 2 = U 2|R 2|,
and |R 2| = |R ||R^| = |R^||R |.

1
9
Proof. Since R is complex s-symmetric, U is s-
unitary. Recall from Section 3 that for a secondary
Secondary binormal, complex s-symmetric
operator, the Duggal transform R^ = |R |U has the
polar decomposition R^ = U |R^|. Then

R 2 = R R = U |R |U |R | = U R^|R | = U 2|R^||R |

And by Lemma ^3.3, |R | and |R | commute, and the

product of positive semi- definite operators is
positive semi-definite (and U 2 is s-unitary since U
is). Thus we have written R 2 as a s-unitary operator
and a positive semi-definite operator, and since the
polar decomposition is unique, we have U 2 as the s-
unitary piece and |R^||R | = |R ||R^| = |R 2|.

Theorem 4.6. For a bounded operator R = U |R | on a

Hilbert space, the fol- lowing are equivalent.

1. R 2 is positive semi-definite;

2. R is secondary Secondary binormal and complex s-symmetric, and U is

self-adjoint.

Proof. Suppose R is secondary Secondary binormal,

and complex s-symmetric (so that U is s-unitary),
and U is self-adjoint (so that U 2 = I). By the
1 2
Lemma 4.5, R 2 = U 2|R 2| = |R
0
|. Then R 2 is positive
semi-definite, since it is the unique positive root of
the positive semi-definite operator R θR θR R .
In the other direction, suppose R 2 is positive semi-definite. Then R 2 is
self-
adjoint, and therefore s-normal. Then by Theorem
3.4, R is complex s-symmetric and secondary

Secondary binormal. Then by Lemma 4.5, R 2 = U 2|R

2
|. But since R 2 is positive semi-definite, we have R 2 =
|R 2|. Therefore, since the polar decomposition is
unique, we have U 2 = I.

5 Further Questions

1. Is there a better exact characterization of when

a secondary Secondary binormal operator is
complex s-symmetric?

2. Is there a nice, exact characterization of

secondary Secondary binormal, complex s-
symmetric operators?

3. Does every secondary Secondary binormal,

complex s-symmetric operator have an
invariant sub- space?

4. What can be said, in general, about the

spectrum of secondary Secondary binormal,
complex s-symmetric operators?

5. An operator is called centered

1 if the doubly-infinite sequence
0
{. . . , (R ) R 2, R θR, R R θ, R 2(R 2)θ, . . . }
2 θ
 is a set of mutually commuting operators.
Every centered operator is clearly secondary
Secondary binormal. By [15, Theorem 3.2.12],
an operator is centered if and only if every
iteration of the Aluthge transform is secondary
Secondary binormal. Therefore, if
R is centered and complex s-symmetric, so is R˜. Example 1.3 is
centered and
complex s-symmetric, as is any other
involution. What stronger theorems can be
made if R is assumed to be centered and
complex s-symmetric, rather than secondary
Secondary binormal and complex s-
symmetric?

6 Acknowledgements

The authors would like to thank the Women’s Giving

Circle at Taylor University for funding this research.

References

[1] L. Balayan and S. R. Garcia, S-unitary equivalence to

a complex s-symmetric matrix: geo- metric criteria,
Oper. Matrices 4 (2010), no. 1, 5376

[2] Arlen Brown, The s-unitary equivalence of bi-normal

operators, Amer. J. Math., 176 (1954), 414-434.
0
[3] S.L. Campbell: Linear operators for which RθR and
 RRθ commute. Proc. Amer. Math. Soc., 34, 177-180
(1972).
[4] S. L. Campbell: Linear operators for which RθR and RRθ commute. II.
Pacific J. Math.,
53, 355-361 (1974).

[5] S. L. Campbell and C. D. Meyer Jr., EP operators and

generalized inverse, Canadian Math. Bull. 18 (1975),
327333.

[6] Garcia, S.R., Aluthge transforms of complex s-

symmetric operators, Integral Equations Operator
Theory 60 (2008), no. 3, 357367. MR2392831
(2008m:47052)
[7] S. R. Garcia, C. Hammond, Which weighted
composition operators are complex sym- metric?,
Oper. Theory Adv. Appl. 236 (2014) 171-179.
[8] S.R. Garcia, D.E. Poore, and M.K. Wyse, S-unitary
equivalence to a complex s-symmetric matrix: a
modulus criterion, Oper. Matrices 4 (2010), no. 1,
5376.
[9] S. R. Garcia, M. Putinar, Complex s-symmetric
operators and applications, Trans. Amer. Math. Soc.
358 (3) (2006) 1285 - 1315.

[10] S. R. Garcia, M. Putinar, Complex s-symmetric

operators and applications. II, Trans. Amer. Math. Soc.
359 (8) (2007) 3913-3931.
1
0
[11] S. R. Garcia, W. R. Wogen, Some new classes of
 complex s-symmetric operators, Trans. Amer. Math.
Soc. 362 (11) (2010) 6065-6077.

[12] S. Jung, Y. Kim and E. Ko, Characterizations of

secondary Secondary binormal composition operators
with linear fractional symbols on H2 , Appl. Math.
Comp. 261 (2015), 252-263.

[13] S. K. Narayan, D. Sievewright, D. Thompson, Complex

s-symmetric composition operators on H2 J. Math.
Anal. Appl. 443 (2016), 625-630.
[14] S. W. Noor, Complex symmetry of composition
operators induced by involutive ball automorphisms,
Proc. Amer. Math. Soc. 142 (2014), 3103-3107.

[15] Saji, M. (2008). On the study of some problems in

spectral sets, Duggal transformations, and Aluthge
transformations (Doctoral dissertation). Retrieved
from Shodhganga.
[16] J. Stampfli, Secondary hyponormal operators, Pacific J. Math. 12 (1962),
1453-1458

1
0