Operators

and
Matrices
Volume 12, Number 4 (2018), 933–942 doi:10.7153/oam-2018-12-56

NONLINEAR MAPS PRESERVING CONDITION SPECTRUM

OF JORDAN SKEW TRIPLE PRODUCT OF OPERATORS

H. B ENBOUZIANE , Y. B OURAMDANE , M. E CH -C HERIF E L K ETTANI

AND A. L AHSSAINI

(Communicated by L. Molnár)

Abstract. Let B(H ) the algebra of all bounded linear operators on a complex Hilbert space
H with dim H > 3 . Let W , V be subsets of B(H ) which contain all rank-one opera-
tors. Denote by rε (A) the condition spectral radius of A ∈ B(H ) . We determine the form of
surjective maps φ : W → V satisfying rε (AB∗ A) = rε (φ (A)φ (B)∗ φ (A)) for all A , B in W ,
we characterize also the structure of surjective maps φ : B(H ) → B(H ) with σε (AB∗ A) =
σε (φ (A)φ (B)∗ φ (A)) for all A , B in B(H ) where σε (A) is the ε -condition spectrum of an
operator A in B(H ) .

1. Introduction

Throughout this note, H will denote a Hilbert space over the complex field C
and B(H ) will denote the algebra of all bounded linear operators on H with unit I .
For A ∈ B(H ), kAk is the norm of A. Let 0 < ε < 1 , the ε -condition spectrum of
A, σε (A), is defined by

σε (A) := {z ∈ C : kz − Akk(z − A)−1k > ε −1 },

with the convention that kz − Akk(z − A)−1k = ∞ when z − A is not invertible, and is
a nonempty compact set contains the usual spectrum σ (A) of A ∈ B(H ), [13]. The
ε -condition spectral radius of A, rε (A), is given by

rε (A) := sup{|z|, z ∈ σε (A)}.

The ε -pseudo spectrum of A, for ε > 0 , Λε (A), is defined by

Λε (A) := {z ∈ C : k(z − A)−1k > ε −1 },

with the convention that k(z − A)−1 k = ∞ when z − A is not invertible. The definition
of the ε -pseudo spectral radius of A, rε′ (A), is given by

rε′ (A) := sup{|z|, z ∈ Λε (A)}.

Mathematics subject classification (2010): Primary 47B49, Secondary 47B48, 47A10, 46H05.
Keywords and phrases: Condition spectra, condition spectral radius, nonlinear Jordan skew triple prod-
uct preservers.

c D l , Zagreb 933
Paper OaM-12-56
934 H. B ENBOUZIANE , Y. B OURAMDANE , M. E CH -C HERIF E L K ETTANI AND A. L AHSSAINI

For more details and basic properties about the condition spectrum and pseudo spectra
we refer the readers to [13, 14, 16]. The inner product on H will be denoted by h., .i.
As usual for x, f ∈ H , x ⊗ f denotes the rank at most one operator on H defined by
z 7→ hz, f ix for every z ∈ H and every operator of rank at most one on B(H ) can be
written in this form. For an operator T ∈ B(H ), T ∗ denotes its adjoint.
The problem of characterizing maps on matrices or operators that preserve cer-
tain functions, subsets and relations has attracted the attention of many mathematicians
in the last decade; for example we can see [1, 2, 3, 4, 5, 6, 9, 10, 11, 14, 15] and their
references. In recent years, a great activity has occurred in characterising maps preserv-
ing condition spectrum, pseudo spectra, condition spectral radius and pseudo spectral
radius. The study of linear case was done by Kumar and Kulkarni in [14]. The non-
linear case has also been studied in recent papers [2, 3, 6, 7, 8], the authors changed
the linearity of such map φ by the weaker condition F(φ (A ⋄ B)) = F(A ⋄ B) where
A ⋄ B is one of different kinds of operations such as sum A + B, the usual product AB,
triple product ABA, the Jordan product AB + BA, Lie product AB − BA and skew prod-
uct A∗ B. F(.) is one of the following objects σε (.), rε (.), Λε (.) or rε′ (.). In [3],
Bendaoud et al. characterised surjective maps preserving the condition spectrum and
condition spectral radius of product and Jordan triple product of operators. In particular
they showed that a surjective map φ from B(H ) into itself, where H is a Hilbert
space with dimH > 3 , satisfies

rε (φ (A)φ (B)φ (A)) = rε (ABA), (A, B ∈ B(H )),

if and only if there exist a functional h : B(H ) → C, with |h(A)| = 1 for all A ∈
B(H ), and a unitary or anti-unitary operator U on H such that either φ (A) =
h(A)UAU ∗ or φ (A) = h(A)UA∗U ∗ for all A ∈ B(H ). They showed also that a sur-
jective map φ from B(H ) into itself, where H is Hilbert space with dim H > 3 ,
satisfies

σε (φ (A)φ (B)φ (A)) = σε (ABA), (A, B ∈ B(H ))

if and only if there is a third root of unity c and a unitary operator U ∈ B(H ) such
that either φ (A) = cUAU ∗ or φ (A) = cUAtrU ∗ for all A ∈ B(H ), where Atr is the
transpose of A with respect to an arbitrary but fixed orthogonal basis of H .
Let W , V be two subsets of B(H ) which contain all rank-one operators. The
aim of this note concern the study of the surjective maps φ : W → V satisfying
• rε (φ (A)φ (B)∗ φ (A)) = rε (AB∗ A) for all A, B ∈ W .
As a consequence we obtain the description of surjective maps φ : B(H ) →
B(H ) with
• σε (φ (A)φ (B)∗ φ (A)) = σε (AB∗ A) for all A, B ∈ B(H ).
The proof of such a promised results uses some arguments that are influenced by
ideas from [2, 8, 10].
 J ORDAN SKEW TRIPLE PRODUCT PRESERVERS 935

2. Preliminaries

In this section, we collect some lemmas that will be used in the proof of our main
results. In the sequel we denote by ε a positive number such that 0 < ε < 1 . We begin
by gathering some properties of the ε -condition spectral radius.

L EMMA 2.1. Let H be a complex Hilbert space and T ∈ B(H ). Then we have

1. rε (T ) = 0 if and only if T = 0 .

2. rε (UTU ∗ ) = rε (T ), for all unitary operators U ∈ B(H ).

3. rε (λ T ) = |λ |rε (T ) for λ ∈ C.

4. rε (T ∗ ) = rε (T ).

Proof. For the first three propositions see [2, Proposition 2.6], the fourth is an
immediate consequence of the definition. 

The second result gives link between rank-one operators which have the same ε -
condition spectral radius in particular case.

L EMMA 2.2. Let u, v ∈ H . If rε (u ⊗ f ) = rε (v ⊗ f ) for every f ∈ H , then the

vectors u and v are linearly dependent.

Proof. See [2, Proposition 2.6]. 

The following lemma characterizes the ε -condition spectrum and the ε -condition
spectral radius of nontrivial projection.

L EMMA 2.3. Let P ∈ B(H ) be a nontrivial projection. Then

1 ε −ε 2 ε
σε (P) = D( , ) ∪ D( , ),
1 − ε2 1 − ε2 1 − ε2 1 − ε2

where D(a, r) is the closed disc centered at a ∈ C with radius r > 0 .

In particular
1
rε (P) = .
1−ε

Proof. See [2, Proposition 2.4] 

We close this section by the following lemma which characterises the maps pre-
serving zero Jordan skew triple products of operators in both directions.
936 H. B ENBOUZIANE , Y. B OURAMDANE , M. E CH -C HERIF E L K ETTANI AND A. L AHSSAINI

L EMMA 2.4. Let H be a complex Hilbert space with dim H > 3 . Let W , V be
subsets of B(H ) which contain all rank-one operators. Suppose that φ : W → V is a
surjective map. Then φ satisfies AB∗ A = 0 ⇔ φ (A)φ (B)∗ φ (A) = 0 for all A, B ∈ W if
and only if there exist unitary or anti-unitary operators U , V on H and a functional
h : W → C \ {0} such that φ is of one of the forms

φ (A) = h(A)UAV f or all A ∈ W ,

or
φ (A) = h(A)UA∗V f or all A ∈ W .

Proof. See [10, Corollary 3.5]. 

3. Main results

We begin by stating and proving the promised main results, since all the necessary
ingredients are collected in the preliminary section. The first theorem characterizes
surjective maps preserving the ε -condition spectral radius of Jordan skew triple product
of operators.

T HEOREM 3.1. Let H be a complex Hilbert space with dim H > 3 . Let W , V
be subsets of B(H ) which contain all rank-one operators. Suppose that φ : W → V
is a surjective map. Then φ satisfies

rε (AB∗ A) = rε (φ (A)φ (B)∗ φ (A)) f or all A, B ∈ W (3.1)

if and only if one of the following statements holds.

1. There exist a unitary or an anti-unitary operator U on H and a functional
ϕ : W → C with |ϕ (A)| = 1 such that φ (A) = ϕ (A)UAU ∗ for all A ∈ W .
2. There exist a unitary or an anti-unitary operator U on H and a functional
ϕ : W → C with |ϕ (A)| = 1 such that φ (A) = ϕ (A)UA∗U ∗ for all A ∈ W .

Proof. According to Lemma 2.1 it is easy to check that if φ takes one of the
forms (1) or (2) then rε (AB∗ A) = rε (φ (A)φ (B)∗ φ (A)) for all A, B ∈ W and so the
sufficiency condition is obvious.
Conversely, assume that φ : W → V is a surjective map satisfying the equation
(3.1). We divide the proof into three steps.
Step 1: φ (A) = h(A)UAV or φ (A) = h(A)UA∗V for every A ∈ W with U ,V
unitary or anti-unitary operators on H and |h(A)| is constant.
For establishing this step we borrow the idea from [10, Theorem 3.1]. Indeed,
since φ satisfies the equation (3.1), it follows from Lemma 2.1 that the map φ preserves
zero Jordan skew triple product of operators in both directions, which means that

AB∗ A = 0 ⇐⇒ φ (A)φ (B)∗ φ (A) = 0 (A, B ∈ W ).

 J ORDAN SKEW TRIPLE PRODUCT PRESERVERS 937

Consequently by Lemma 2.4 there exist a unitary or an anti-unitary operators U , V on

H and a functional h : W → C \ {0} such that either φ (A) = h(A)UAV or φ (A) =
h(A)UA∗V for all A ∈ W . It remains to prove that |h(A)| is constant for every A ∈ W
in the both cases.
• Assume that φ (A) = h(A)UAV for all A ∈ W . It follows by equation (3.1) that
for every A, B in W , rε (AB∗ A) =| h(A)2 h(B) | rε (UAB∗ AV ). In particular if A = x ⊗ f
for x and f in H with hB∗ x, f i =
6 0 , we get
rε (x ⊗ f B∗ x ⊗ f ) =| h(x ⊗ f )2 h(B) | rε (Ux ⊗ f B∗x ⊗ fV ),
which is equivalent to
| hB∗ x, f i | rε (x ⊗ f ) =| h(x ⊗ f )2 h(B)hB∗ x, f i | rε (Ux ⊗ V ∗ f ),
thus
rε (x ⊗ f )
|h(B)| = .
|h(x ⊗ f )2 |rε (Ux ⊗ V ∗ f )
Since for any non-zero B1 , B2 ∈ W , there exist x , f ∈ H such that hB∗i x, f i = 6 0,
i = 1, 2 , one gets |h(B1 )| = |h(B2 )|. Finally we conclude that there exists a constant
β > 0 such that |h(A)| = β for all A ∈ W .
• Assume that φ (A) = h(A)UA∗V for all A ∈ W . It follows by equation (3.1) that
rε (AB∗ A) =| h(A)2 h(B) | rε (UA∗ BA∗V ) for all A, B in W . In particular if A = x ⊗ f
for x and f in H with hB∗ x, f i = 6 0 , we have
rε (x ⊗ f B∗ x ⊗ f ) =| h(x ⊗ f )2 h(B) | rε (V ∗ (x ⊗ f )B∗ (x ⊗ f )U ∗ ),
which is equivalent to
| hB∗ x, f i | rε (x ⊗ f ) =| h(x ⊗ f )2 h(B)hB∗ x, f i | rε (V ∗ x ⊗ U f ),
thus
rε (x ⊗ f )
|h(B)| = .
|h(x ⊗ f )2 |rε (V ∗ x ⊗ U f )
With the similar reasoning as above one gets |h(A)| = β , for all A ∈ W with β being
a positive constant.
Step 2: U = α V ∗ for some nonzero scalar α .
For establishing this point we will discuss the forms given in the previous step.
Note that |h(A)| = β is constant in both cases for all A ∈ W .
• Assume that φ (A) = h(A)UAV for every A ∈ W .
Let x ∈ H be a nonzero vector, then for every nonzero vector f ∈ H , we have
rε (x ⊗ f (x ⊗ f )∗ x ⊗ f ) = rε (x ⊗ f ( f ⊗ x)x ⊗ f )
= kxk2 k f k2 rε (x ⊗ f ).
On the other hand we have
rε (φ (x ⊗ f )(φ (x ⊗ f ))∗ φ (x ⊗ f )) = β 3 rε (Ux ⊗ fVV ∗ ( f ⊗ x)U ∗Ux ⊗ fV )
= β 3 kxk2 k f k2 rε (Ux ⊗ fV )
= β 3 kxk2 k f k2 rε (VUx ⊗ f ).
938 H. B ENBOUZIANE , Y. B OURAMDANE , M. E CH -C HERIF E L K ETTANI AND A. L AHSSAINI

Combining this with equation (3.1), we get β 3 kxk2 k f k2 rε (VUx⊗ f ) = kxk2 k f k2 rε (x⊗
f ), so rε (x ⊗ f ) = β 3 rε (VUx ⊗ f ), thus by Lemma 2.2, x and VUx are linearly de-
pendent for every x ∈ H . We conclude that VU = α I for some scalar α .
• Assume that φ (A) = h(A)UA∗V for every A ∈ W .
Note that rε (T ) = rε (T ∗ ) for every T ∈ B(H ).
Let f ∈ H be a nonzero vector, then for every nonzero vector x ∈ H , we have
rε (x ⊗ f (x ⊗ f )∗ x ⊗ f ) = rε (x ⊗ f ( f ⊗ x)x ⊗ f )
= kxk2 k f k2 rε (x ⊗ f )
= kxk2 k f k2 rε ( f ⊗ x).
On the other hand
rε (φ (x ⊗ f )(φ (x ⊗ f ))∗ φ (x ⊗ f )) = β 3 rε (U f ⊗ xVV ∗ (x ⊗ f )U ∗U f ⊗ xV )
= β 3 kxk2 k f k2 rε (U f ⊗ xV )
= β 3 kxk2 k f k2 rε (VU f ⊗ x).
As before we have rε ( f ⊗ x) = β 3 rε (VU f ⊗ x). It follows by Lemma 2.2 that f
and VU f are linearly dependent for every f ∈ H . This implies that VU = α I for
some scalar α .
Step 3: φ takes the desired form.
Recall that we already obtained that φ (A) = ϕ (A)UAU ∗ or φ (A) = ϕ (A)UA∗U ∗
for every A ∈ W , with ϕ (A) = α h(A) and |ϕ (A)| is constant for every A ∈ W . It
remains to show that |ϕ (A)| = 1 for every A ∈ W . Indeed, let x ∈ H be a unit vector.
If φ (A) = ϕ (A)UAU ∗ for every A ∈ W , then by equation (3.1), we have
rε (x ⊗ x) = rε (x ⊗ x(x ⊗ x)∗x ⊗ x)
= rε (φ (x ⊗ x)φ (x ⊗ x)∗ φ (x ⊗ x))
= |ϕ (x ⊗ x)|3 rε (Ux ⊗ xU ∗)
= |ϕ (x ⊗ x)|3 rε (x ⊗ x).
1
By Lemma 2.3, rε (x ⊗ x) = , so it follows that |ϕ (x ⊗ x)|3 = 1 . Thus |ϕ (A)| = 1
1−ε
for every A ∈ W .
If φ (A) = ϕ (A)UA∗U ∗ for every A ∈ W then
rε (x ⊗ x) = rε (x ⊗ x(x ⊗ x)∗x ⊗ x)
= rε (φ (x ⊗ x)φ (x ⊗ x)∗ φ (x ⊗ x))
= |ϕ (x ⊗ x)|3 rε (Ux ⊗ xU ∗)
= |ϕ (x ⊗ x)|3 rε (x ⊗ x).
1
Again by Lemma 2.3, rε (x ⊗ x) = which implies that |ϕ (x ⊗ x)|3 = 1 and thus
1−ε
|ϕ (A)| = 1 for every A ∈ W , which completes the proof. 
Now we are able to characterize surjective maps on B(H ) preserving ε -condition
spectrum of the Jordan skew triple product of operators.
 J ORDAN SKEW TRIPLE PRODUCT PRESERVERS 939

T HEOREM 3.2. Let H be a complex Hilbert space with dim H > 3 . Let φ :
B(H ) → B(H ) be a surjective map. Then φ satisfies

σε (AB∗ A) = σε (φ (A)φ (B)∗ φ (A)) f or all A, B ∈ B(H ) (3.2)

if and only if one of the following statements holds.

1. There exists a unitary operator U ∈ B(H ) such that φ (A) = UAU ∗ for all
A ∈ B(H ).
2. There exists an anti-unitary operator U on H such that φ (A) = UA∗U ∗ for all
A ∈ B(H ).
To prove this theorem, we need the following lemma summarizing some prop-
erties of the ε -condition spectrum proved in [13].

L EMMA 3.3. Let H be a complex Hilbert space and T ∈ B(H ). Then we have
1. σε (α + β A) = α + β σε (T ) for all α , β ∈ C.
2. σε (UTU ∗ ) = σε (T ), for all unitary U ∈ B(H ).
3. σε (UTU ∗ ) = σε (T ), for all anti-unitary U on H .
4. σε (T ∗ ) = σε (T ).
5. σε (I) = {1} .
Proof of Theorem 3.2. The “if“ part is easily verified according to the previous lemma.
To prove the “only if“ part, suppose that φ satisfies the equation (3.2). It follows that
φ preserves the ε -condition spectral radius of Jordan skew triple product of operators.
Then by Theorem (3.1) there exist a unitary or an anti-unitary operator U on H and
ϕ : W → C with |ϕ (A)| = 1 such that either

φ (A) = ϕ (A)UAU ∗ f or all A ∈ B(H ), (3.3)

or
φ (A) = ϕ (A)UA∗U ∗ f or all A ∈ B(H ). (3.4)
We will complete the proof by discussing two cases.
Case 1. φ takes the form (3.3).
• Suppose that U is unitary. Let us prove that ϕ (A) = 1 for every A ∈ B(H ).
We will proceed as in the proof of [10, Corollary 5.4]. Indeed, if φ (A) = ϕ (A)UAU ∗ for
all A ∈ B(H ) with U unitary. It follows from equation (3.2) that for all A ∈ B(H )
we have

σε (A) = σε (φ (I)φ (A∗ )∗ φ (I))

= σε (ϕ (I)2 ϕ (A∗ )UAU ∗ )
= ϕ (I)2 ϕ (A∗ )σε (A).
940 H. B ENBOUZIANE , Y. B OURAMDANE , M. E CH -C HERIF E L K ETTANI AND A. L AHSSAINI

In particular for A = I , since σε (I) = {1} , we have ϕ (I)2 ϕ (I) = 1 , thus ϕ (I) = 1 . So

σε (A) = ϕ (A∗ )σε (A) f or all A ∈ B(H ).

Which implies that, for every A ∈ B(H ),

z ∈ σε (A) ⇔ ϕ (A∗ )z ∈ σε (A).

If σε (A) is not a disc centred at zero, we deduce that ϕ (A∗ ) = 1 . Therefore ϕ (A) = 1
for every A ∈ B(H ). Otherwise if σε (A) is a disc centred at zero. Choose x, f in H
6 0 and hA∗ x, f i =
such that hx, f i = 6 0 . By Lemma 2.3, σε (x ⊗ f ) is not a disc centred
at zero, then by the above consideration, ϕ (x ⊗ f ) = 1 . Furthermore by equation (3.2)

σε (x ⊗ f A∗ x ⊗ f ) = σε (φ (x ⊗ f )φ (A)∗ φ (x ⊗ f ))
= ϕ (x ⊗ f )2 ϕ (A)σε (x ⊗ f A∗ x ⊗ f )
= ϕ (A)hA∗ x, f iσε (x ⊗ f ).

On the other hand σε (x ⊗ f A∗ x ⊗ f ) = hA∗ x, f iσε (x ⊗ f ). Thus for all A ∈ B(H )

hA∗ x, f iσε (x ⊗ f ) = ϕ (A)hA∗ x, f iσε (x ⊗ f ). Since hA∗ x, f i =
6 0 and σε (x ⊗ f ) is not
a circular disc centred at zero, we deduce that ϕ (A) = 1 . Finally ϕ (A) = 1 for all
A ∈ B(H ), and so φ (A) = UAU ∗ for all A ∈ B(H ).
• Assume that U is anti-unitary.
Let us show that this case cannot occur. We will use an idea given in [2, Theorem
3.5]. Assume that φ (A) = ϕ (A)UAU ∗ for all A ∈ B(H ) and U is an anti-unitary
operator. Note that by Lemma 2.3, one can see that for every unit vector x ∈ H

σε (x ⊗ x) = σε (x ⊗ x).

Following a similar discussion as in the previous case we get that for every A ∈ B(H ),

σε (A) = ϕ (A)ϕ (A∗ )σε (A).

In particular for A = I we get ϕ (I) = 1 . Moreover if x ∈ H is a unit vector, by

equation (3.2), we get

σε (x ⊗ x) = σε (x ⊗ xIx ⊗ x))
= σε (ϕ (x ⊗ x)2 ϕ (I)Ux ⊗ xU ∗)
= ϕ (x ⊗ x)2 σε (x ⊗ x)
= ϕ (x ⊗ x)2 σε (x ⊗ x).

Thus ϕ (x ⊗ x)2 = 1. Now take an arbitrary non self adjoint operator A ∈ B(H ) and a
unit vector x ∈ H such that hA∗ x, xi =
6 0 . Then it follows that

σε (x ⊗ xA∗ x ⊗ x) = σε (φ (x ⊗ x)φ (A)∗ φ (x ⊗ x))

= ϕ (x ⊗ x)2 ϕ (A)σε (Ux ⊗ xA∗ x ⊗ xU ∗)
= ϕ (A)hA∗ x, xiσε (x ⊗ x).
 J ORDAN SKEW TRIPLE PRODUCT PRESERVERS 941

This implies that

ϕ (A)hA∗ x, xi = hA∗ x, xi,
and so
hA∗ x, xi
= ϕ (A).
hA∗ x, xi
In particular for A ∈ B(H ) for which there exist nonzero vectors x, y ∈ H satisfying
A∗ x = x and A∗ y = iy we get

hA∗ x, xi hA∗ y, yi
1= = ϕ (A) = = −1,
hA∗ x, xi hA∗ y, yi

which is a contradiction.
Case 2. φ takes the form (3.4).
We have φ (A) = ϕ (A)UA∗U ∗ for all A ∈ B(H ). With similar discussion as in
the first case with U unitary or anti-unitary we deduce that ϕ (I) = 1.
• Let U be unitary. By equation (3.2) and the fact that ϕ (I) = 1 , we have

σε (A) = ϕ (A∗ )σε (A∗ ) for all A ∈ B(H ).

Then following the same steps as the second discussion in the first case we deduce that
this case cannot occur.
• Assume that U is anti-unitary. To complete the proof let us show that ϕ (A) = 1
for every A ∈ B(H ). We have φ (A) = ϕ (A)UA∗U ∗ for every A ∈ B(H ). It follows
that σε (A) = ϕ (A∗ )σε (A). As in the above consideration we conclude that ϕ (A) = 1
for all A ∈ B(H ). Then φ takes the desired form.

Acknowledgements. the authors would to thank warmly the Professor Lajos Mol-
nar and the anonymous referee for carefully reading the paper and for valuable sugges-
tions.

REFERENCES

[1] H. B ENBOUZIANE , Y. B OURAMDANE , M. E CH -C HERIF E L K ETTANI , A. L AHSSAINI , Nonlinear

commutant preservers, Linear and Multilinear Algebra, (2017), doi:10.1080/03081087.2017.
1307915.
[2] M. B ENDAOUD , A. B ENYOUNESS , M. S ARIH, Condition spectra of special operators and condi-
tion spectra preservers, J. Math. Anal. Appl. (2017), http://dx.doi.org/10.1016/j.jmaa.
2016.12.022.
[3] M. B ENDAOUD , A. B ENYOUNESS , M. S ARIH, Preservers of pseudo spectral radius of operator
products, Linear Algebra Appl. 489 (2016) 186–198.
[4] R. B HATIA , P. Š EMRL , A. R. S OUROUR, Maps on matrices that preserve the spectral radius distance,
Studia Math. 134 (1999) 99–110.
[5] A. B OURHIM , T. JARI , J. M ASHREGHI , Peripheral local spectrum preservers and maps increasing
the local spectral radius, Oper. Matrices 1 (2016) 189–208.
[6] J. C UI , V. F ORSTALL , C. K. L I , V. YANNELLO, Properties and preservers of the pseudospectrum,
Linear Algebra Appl. 436 (2012) 316–325.
[7] J. C UI , C. K. L I , Y. T. P OON, Pseudospectra of special operators and pseudospectrum preservers, J.
Math. Anal. Appl. 419 (2014) 1261–1273.
942 H. B ENBOUZIANE , Y. B OURAMDANE , M. E CH -C HERIF E L K ETTANI AND A. L AHSSAINI

[8] J. C UI , C. K. L I , AND N. S. S ZE, Unitary similarity invariant function preservers of skew products
of operators, J. Math. Anal. Appl. (2017), http://dx.doi.org/10.1016/j.jmaa.2017.04.072.
[9] M. E CH - CH ÉRIF E L K ETTANI AND H. B ENBOUZIANE, Additive maps preserving operators of inner
local spectral radius zero, Rendiconti del Circolo Matematico di Palermo 63 (2) (2014) 311–316.
[10] J. H OU , K. H E , X. Z HANG, Nonlinear maps preserving numerical radius of indefinite skew products
of operators, Linear Algebra Appl. 430 (2009) 2240–2253.
[11] A. A. JAFARIAN , A. R. S OUROUR, Linear maps that preserve the commutant, double commutant or
the lattice of invariant subspaces, Linear and Multilinear Algebra. 38 (1994) 117–129.
[12] A. K RISHNAN AND S. H. K ULKARNI , Pseudo spectrum of element in Banach algebra, Oper. Matrices
1 (2017) 263–287.
[13] S. H. K ULKARNI , D. S UKUMAR, The condition spectrum, Acta Sci. Math. (Szeged) 74 (2008) 625–
641.
[14] G. K. K UMAR , S. H. K ULKARNI , Linear maps preserving pseudospectrum and condition spectrum,
Banach J. Math. Anal. 6 (2012) 45–60.
[15] C. K. L I , P. Š EMRL , N. K. T SING, Maps preserving the nilpotency of products of operators, Linear
Algebra Appl. 424 (2007) 222–239.
[16] L. N. T REFETHEN , M. E MBREE, Spectra and Pseudospectra: The Behavior of Nonnormal Matrices
and Operators, Princeton University Press, Princeton, 2005.

(Received August 3, 2017) H. Benbouziane

Department of Mathematics
Faculty of Sciences DharMahraz Fes
University Sidi Mohammed Ben Abdellah
1796 Atlas Fes, Morocco
Y. Bouramdane
Department of Mathematics
Faculty of Sciences DharMahraz Fes
University Sidi Mohammed Ben Abdellah
1796 Atlas Fes, Morocco
M. Ech-Cherif El Kettani
Department of Mathematics
Faculty of Sciences DharMahraz Fes
University Sidi Mohammed Ben Abdellah
1796 Atlas Fes, Morocco
A. Lahssaini
Department of Mathematics
Faculty of Sciences DharMahraz Fes
University Sidi Mohammed Ben Abdellah
1796 Atlas Fes, Morocco

Operators and Matrices

www.ele-math.com
[email protected]