Abstract
Let X be an infinite-dimensional Banach space, and \(\mathcal {B}(X)\) be the algebra of all bounded linear operators on X. A map \(\Delta \), from \( \mathcal {B}(X) \) into a closed subsets of \( \mathbb {C}\ \) is said to be \( \partial \)-spectrum if \( \partial (\sigma (T))\subseteq \Delta (T) \subseteq \sigma (T) \) for all \(T \in \mathcal {B}(X)\). Here, \( \sigma (T) \) is spectrum of T and \( \partial (\sigma (T)) \) the boundary of \(\sigma (T)\). In this paper, we determine the forms of all surjective maps \( \phi \) from \( \mathcal {B}(X) \) into itself that satisfy either \( \Delta (\phi (T)\phi (S)) = \Delta ( TS) \) for all \(T, S \in \mathcal {B}(X)\) or \( \Delta (\phi (T)\phi (S)\phi (T)) = \Delta (TST) \) for all \( T, S\in \mathcal {B}(X) \).
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction and Preliminaries
Let X be an infinite-dimensional Banach space with dual space \( X^{*} \) and \(\mathcal {B}(X)\) be the algebra of all bounded linear operators on X. The identity operator on X (resp \( X^{*} \)) will denote by I (resp \( I_{X^{*}} \)).
For \( x\in X{\setminus }\{0\} \) and \( f \in X^{*}{\setminus }\{0\} \), we denote by \( x \otimes f \) the bounded linear rank one operator defined by \( (x \otimes f)y = f(y)x \) for all \(y \in X \). Note that every operator on X of rank one can be written as \( x \otimes f \) for some \( x \in X{\setminus }\{0\}\) and \( x \in X^{*}{\setminus }\{0\}\). Note that \( \sigma (x\otimes f)=\{0,f(x)\} \), and that \( x\otimes f \) is a nilpotent operator if and only if \( f(x) = 0 \). The set of all rank one operators, the set of all nilpotent operators and the ideal of all finite rank operators in \(\mathcal {B}(X)\) are denoted by \(\mathcal {F}_{1}(X)\), \(\mathcal {N}_{1}(X)\) and \(\mathcal {F}(X)\), respectively.
For \( T \in \mathcal {B}(X)\), we denote by R(T), N(T), \(T^{*}\), \( \sigma (T) \), \( \sigma _{l}(T)\), \( \sigma _{r}(T )\), \( \sigma _{sur}(T)\), \( \sigma _{ap}(T) \) and \( \partial (\sigma (T)) \), the the range, the kernel, the adjoint, the spectrum, the left spectrum, the right spectrum, the surjectivity spectrum, the approximate point spectrum and the boundary of spectrum of T, respectively. The hyper-range of \(T \in \mathcal {B}(X)\) is defined by \( T^{\infty }(X):=\bigcap _{n\in \mathbb {N}}T(X^n)\).
Consider the map \(\Delta : \mathcal {B}(X) \rightarrow \{\text {closed set of } \mathbb {C}\} \) with \( \Delta (.) \) is any one of the spectral functions \(\sigma (.),~\sigma _{l}(.),~ \sigma _{r}(.),~ \sigma _{ap}(.)\) or \(\sigma _{sur}(.) \). It is known that these spectral functions satisfy
In general, a map \(\Delta : \mathcal {B}(X) \rightarrow \{\text {closed subsets of } \mathbb {C}\} \) is said to be \(\partial \)-spectrum if
for all \(T \in \mathcal {B}(X)\); see [15, Def 3.2]. Note that if \(\Delta \) is \(\partial \)-spectrum, then for every \(T\in \mathcal {B}(X)\), \(\Delta (T)\) is non-empty and
and in this case, we have \(\Delta (T)= \sigma (T)\). In particular,
An operator \( T \in \mathcal {B}(X)\) is called Kato operator (or semi-regular) if R(T) is closed and \( N(T)\subseteq T^{\infty }(X) \), the set \( \sigma _{K}(T)=\{\lambda \in \mathbb {C}/ T-\lambda I\) is not kato operator} is called the Kato spectrum. Mbekhta and Ouahab [16] proved that
for all \(T \in \mathcal {B}(X)\), and thus \(\sigma _{K}(.) \) is \( \partial \)-spectrum.
Note that there are other spectra satisfying property (1.1), namely the generalized spectrum \(\sigma _{g}(T)\) of operator T, the Saphar spectrum \(\sigma _{rr}(T)\). For more information about these spectra, we refer the reader to [13, 19].
The problem of characterizing maps on matrices or operators that preserve certain functions, subsets and relations has attracted the attention of many mathematicians in the last decade; see for example [1, 4,5,6, 8, 11, 12, 14, 15, 20, 22].
In [6], Cui and Hou showed that if \(\Delta \) is a \( \partial \)-spectrum, and \( \phi \) is a linear map from a semisimple Banach algebra \(\mathcal {A}\) onto another one \(\mathcal {B}\) such that
then \(\phi \) is idempotent preserving and \(\phi (I)=I\).
In the last decades, many authors, investigated maps preserving a certain property of the product or triple product without assuming linearity or additivity. We refer the interested reader to [2,3,4, 9, 10, 17, 18, 23].
In [18], Molnár described maps preserving the spectrum of product of operators. In particular, he showed that a surjective map \(\phi : \mathcal {B}(H)\rightarrow \mathcal {B}(H)\) (H is an infinite-dimensional complex Hilbert space) satisfies
if and only if there exists an operator invertible \(A\in \mathcal {B}(H)\) such that either \(\phi (T)=ATA^{-1}\) for all \(T\in \mathcal {B}(H)\) or \(\phi (T)=-ATA^{-1}\) for all \(T\in \mathcal {B}(H)\). He obtained, in the same paper, a similar result by considering the surjectivity spectrum and the point spectrum instead the usual spectrum.
Instead of the usual product, certain authors investigated maps preserving spectra of triple product of operators or matrices; See for instance [7, 9, 23]. In [23], Zhang and Hou gave the form of maps \(\phi : \mathcal {B}(X)\rightarrow \mathcal {B}(X)\) satisfying
where \(\sigma _{\pi }(T):=\{\lambda \in \sigma (T) / \mid \lambda \mid =r(T)\}\) is the peripheral spectrum of \(T \in \mathcal {B}(H)\).
Our objective of this paper is to study nonlinear maps preserving any part of the spectrum, which contains the boundary of the spectrum (\(\partial \)-spectrum), of the product of operator. We thus obtain Molnár’s main results in the case of an arbitrary Banach space using a different approach. We also obtain the form of any surjective map preserving the \(\partial \)-spectrum of the triple product of operators. Our proofs are inspired by those of the main results of the papers [2, 3].
Throughout this paper, let \(\Delta \) be a \(\partial \)-spectrum map.
2 Maps Preserving the \( \partial \)-Spectrum of Product of Operators
The following theorem is our main result in this section which characterizes nonlinear maps preserving a \( \partial \)-spectrum \(\Delta (.)\) of the product of operators.
Theorem 2.1
Let \(\phi :\mathcal {B}(X)\rightarrow \mathcal {B}(X)\) be a surjective map satisfying
then there exists a scalar \( \alpha =\pm 1 \) and either there is a bounded invertible operator \( A: X \rightarrow X \) such that
or there is a bounded invertible operator \( C: X^{*} \rightarrow X \) such that
In the last case, X is automatically reflexive.
Before embarking on the proof, we need several lemmas. The first one was proved in [20].
Lemma 2.2
[20, Theorem 3.3] Let \( \phi : \mathcal {F}(X) \rightarrow \mathcal {F}(X)\) be a bijective linear map preserving rank one operators in both directions. Then, either there are linear bijective mappings \( A: X \rightarrow X \) or \( B: X^{*} \rightarrow X^{*} \) such that
or, there are linear bijective mappings \( C: X^{*} \rightarrow X \) and \( D: X \rightarrow X^{*} \) such that
For the remaining lemmas and the proofs of the main theorems, we will need the following notation—
where \( \Delta \) is \(\partial \)-spectrum and \( T \in \mathcal {B}(X) \). In particular, we have \( \Delta ^{*}(x\otimes f)=\{f(x)\} \) for all \( x \in X \) and \( f \in X^{*} \).
The next lemma gives necessary and sufficient conditions for two operators to be equal in term of \( \partial \)-spectrum.
Lemma 2.3
Let \( T,~ S~\in \mathcal {B}(X) \). Then the following statements are equivalent.
-
1.
\( T=S \).
-
2.
\( \Delta (TR) = \Delta (SR)\) for all \( R \in \mathcal {F}_{1}(X) \).
-
3.
\( \Delta ^{*}(TR) = \Delta ^{*}(SR)\) for all \( R \in \mathcal {F}_{1}(X) \).
Proof
We only need to prove that the implication \((3)\Rightarrow (1)\) holds.
Assume that \( \Delta ^{*}(TR) = \Delta ^{*}(SR)\) for all \( R \in \mathcal {F}_{1}(X) \) and let \(x \in X\) and \( f\in X^{*}\). We have
and thus \( Tx=Sx \). By the arbitrariness of x, clearly \( T=S \).
\(\square \)
The following lemma gives a characterization of rank one operators in term of \( \partial -\) spectrum.
Lemma 2.4
Let \( R \in \mathcal {B}(X)\setminus \{0\}\). The following statements are equivalent:
-
1.
R has rank one.
-
2.
\( \Delta ^{*}(TR \)) is a singleton for all operator \( T \in \mathcal {B}(X) \).
Proof
Note that, for any \(T \in \mathcal {B}(X)\), if \( \Delta ^{*}(T)\) is a singleton, then \( \Delta ^{*}(T)= \sigma _{\pi }(T)\). Thus, this Lemma is an immediate consequence of [17, Lemma 2.1].
\(\square \)
Lemma 2.5
Let \( T,S \in \mathcal {B}(X)\). Then for every \(R \in \mathcal {F}_{1}(X)\), we have
\( \Delta ^{*}((T+S)R) =\Delta ^{*}(TR)+ \Delta ^{*}(SR)\).
Proof
Let \(R \in \mathcal {F}_{1}(X)\) such that \(R=x\otimes f\) where \(x \in X, ~ f\in X^{*} \). Note that \( TR=Tx\otimes f \) and \( SR=Sx\otimes f \). Then
\(\square \)
Proof of Theorem 2. 1.
We will prove the theorem in five steps—
Step 1. \( \phi \) is injective, \( \phi ^{-1} \) also satisfies (2.1) and \( \phi (0)= 0\).
Let \(A,B \in \mathcal {B}(X)\) such that \( \Phi (A) = \Phi (B) \). For every \(R \in \mathcal {F}_{1}(X)\), we have
By Lemma 2.3, we get \( A = B \) which proves that \( \phi \) is injective. It follows that \( \phi \) is a bijection and it is obvious that \( \phi ^{-1} \) satisfies (2.1).
For any \(T \in \mathcal {B}(X)\), we have
Since \(\phi \) is surjective, lemma 2.3 implies that \( \phi (0) =0\).
Step 2. \(\phi \) preserves rank one operators in both directions.
Let \(R \in \mathcal {F}_{1}(X)\) such that \(R=x\otimes f\) where \(x \in X\) and \( f\in X^{*}\). For every \(T \in \mathcal {B}(X)\) there exists \(S\in \mathcal {B}(X)\) such that \(T = \phi (S) \). We have
Then \( \Delta ^{*}(\phi (R)T) \) has one element for all \(T \in \mathcal {B}(X)\). By Lemma 2.4, we see that \( \phi (R)\in \mathcal {F}_{1}(X) \). In addition, since \( \phi \) is bijective and \( \phi ^{-1} \) satisfies (2.1), \( \phi \) preserves rank one operators in the both direction.
Note also that \(\phi \) preserves non-nilpotent rank one operators in both directions. Indeed, let \(R=x\otimes f\) where \(x \in X\) and \(f\in X^{*} \) with \( f(x)\ne 0\). Then \(\phi (R)=y\otimes g\) where \(y \in X\) and \( g\in X^{*}\). On the other hand, we have
Since \( f(x)\ne 0 \), then \( g(y) \ne 0 \). It follows that \( \phi (R) \in \mathcal {F}_{1}(X){\setminus }\mathcal {N}_{1}(X)\). The reverse direction is obvious since \(\phi \) is bijective.
Step 3. \( \phi \) is a linear map preserving finite rank operators in both directions.
Let \(T, S \in \mathcal {B}(X)\). For every \( R \in \mathcal {F}_{1}(X)\) we have
Using Lemma 2.3, we conclude that \( \phi (T+S) = \phi (T) + \phi (S) \). Let \( \lambda \in \mathbb {C} \) and \( T \in \mathcal {B}(X) \), let us prove that \( \phi (\lambda T) = \lambda \phi (T) \).
If \( \lambda = 0 \), Then \( \phi (0T) = \phi (0) = 0 = 0\phi (T) \).
If \( \lambda \ne 0 \), let \(R \in \mathcal {F}_{1}(X) \),
By Lemma 2.3 and the surjectivity of \( \phi \), we conclude that \( \phi (\lambda T) = \lambda \phi (T)\), which proves that \( \phi \) is linear. Finally, since every \(T \in \mathcal {F}(X) \) is a finite linear combination of rank one operators and \( \phi \) preserves \( \mathcal {F}_{1}(X) \) in both directions, by linearity we find that \( \phi \) preserves finite rank operators in both directions.
Step 4. \( \phi (I)=\alpha I \) where \( \alpha ^{2}=1\).
Suppose, by the way of contradiction, that \( \Phi (I) \) and I are linearly independent, then there exists a nonzero vector \(x \in X\) such that \( \phi (I)x \) and x are linearly independent. Let \(f \in X^{*} \) such that \( f(x) = 1 \) and \( f(\phi (I)x) = 0 \). For \( R = x\otimes f \in \mathcal {F}_{1}(X){\setminus }\mathcal {N}_{1}(X) \), by step 2 and surjectivity of \( \phi \), there is \(T \in \mathcal {F}_{1}(X){\setminus }\mathcal {N}_{1}(X) \) such that \( \phi (T) = R \) and \(T = y\otimes g \) where \(y \in \) X and \(g \in X^{*} \) with \( g(y) \ne 0 \). Therefore,
Thus \(g(y)=0\). This contradiction shows that \( \Phi (I)=\alpha I \) for some non-zero scalar \( \alpha \in \mathbb {C}\).
On the other hand, we have
Which implies that \(\alpha ^{2}=1\); as desired.
Step 5. \( \phi \) has the desired form.
Let \( \varphi :=\alpha ^{-1}\phi \), note that \(\varphi \) satisfies (2.1) and \( \varphi (I)=I\). It follows, from Step 3 and Lemma 2.2, that \(\phi \) takes either form (2.4) or form (2.5).
Suppose that (2.4) holds. Let \( x\in X \) and \( f \in X^{*}\), and note that
Then
Now, we are ready to prove that A is bounded and \( B = (A^{*})^{-1} \).
Let \( (x_{n})_{n} \) be a sequence of elements of X and \( x,y \in A \) such that \( \lim \nolimits _{n\rightarrow \infty }x_{n} = x \) and \( \lim \nolimits _{n\rightarrow \infty } Ax_{n} = y \). We show that \( Ax = y \).
For every \( f \in X^{*} \), we have
Since B is bijective, then \( f(Ax) = f(y \)) for all \( x \in X \) and all \( f \in X^{*}\). By Hahn–Banach theorem, we conclude that \( Ax = y \). The closed graph theorem shows that A is bounded. Moreover,
Therefore, for every \(x \in X \) and \( f \in X^{*}\) we have
Then
Now, let \( T \in \mathcal {B}(X )\) and \( R \in \mathcal {F}_{1}(X) \), we have
Since \( \varphi \) is surjective, by Lemma 2.3, we conclude that
Finally, \( \phi (T) = \alpha ATA^{-1} \) with \( \alpha ^{2} = 1 \) for all \( T \in \mathcal {B}(X )\).
Now, suppose that \( \varphi \) takes the second form (2.5). Let \( f \in X^{*} \) and \( x \in X\), we have
This shows that
Consequently, by similar reasoning as the first case and using the closed graph theorem, separately for each of C and D, we conclude that these operators are bounded. Therefore, both \( C^{*}: X^{*} \rightarrow X^{**} \) and \( D^{*}: X^{**} \rightarrow X^{*}\) are invertible. Moreover, if \( \jmath \) is the canonical embedding of X in \( X^{**} \), then we have \( D^{*}\circ \jmath \circ C = I_{X^{*}} \) which implies that \( \jmath \circ C = (D^{*})^{-1 } \). Since both C and \( (D^{*})^{-1 } \) are surjective, \( \jmath \) is also surjective and hence X is reflexive. By identifying X with \( X^{**} \), we conclude that \( D^{*}\circ C = I_{X^{*}} \) and \( C^{*}\circ D = I \). Which implies that \( D=(C^{*})^{-1}=(C^{-1})^{*} \) and
Let \( T \in \mathcal {B}(X)\). Then for every \( f \in X^{*} \) and \( x \in X \), we obtain
By identifying X with \( X^{**} \), we get \( ((C^{-1})^{*}x)(\varphi (T)(Cf))=f(Tx) \). Thus, for every \( f \in X^{*} \) and \( x \in X \), we have \( C^{-1}\varphi (T)Cf(x)=T^{*}f(x) \). Therefore,
Finally, we get \(\phi (T)=\alpha CT^{*}C^{-1} \) with \( \alpha ^{2}=1\), and the proof is thus complete.\(\square \)
3 Surjective Maps Preserving the \( \partial \)-Spectrum of Triple Product of Operators
In this section, we will study maps (without assuming linearity or additivity) preserving the \( \partial \)-spectrum of triple product of operators. We begin this section with the following identity principal which will be used frequently in the proof of Theorem 3.4.
Lemma 3.1
If \( T,~ S~\in \mathcal {B}(X) \), then the following statements are equivalent.
-
1.
\( T=S \).
-
2.
\( \Delta (RTR) = \Delta (RSR)\) for all \( R \in \mathcal {F}_{1}(X)\setminus \mathcal {N}_{1}(X) \).
-
3.
\( \Delta ^{*}(RTR) = \Delta ^{*}(RSR)\) for all \( R \in \mathcal {F}_{1}(X)\setminus \mathcal {N}_{1}(X) \).
Proof
We only need to prove that \((3\implies 1)\). Assume that \( \Delta ^{*}(RTR) = \Delta ^{*}(RSR)\) for all \( R \in \mathcal {F}_{1}(X)\setminus \mathcal {N}_{1}(X) \).
Let \(R=x\otimes f\), where \(x \in X{\setminus }\{0\}\) and \(f\in X^{*}\). If \(f(x)\ne 0 \), note that \( RTR=f(Tx)x\otimes f \) and \( RSR=f(Sx)x\otimes f\). Then
Now, if \( f(x) = 0\), let \( g \in X^{*} \) such that \( g(x) \ne 0 \). Then by the first case
Then
Which implies that \( f(Tx)) = f(Sx) \) in this case too. Therefore, by Hahn-Banach theorem, \(Tx = Sx \) for all \(x \in X \), which proves that \(T =S\). \(\square \)
The second result provides necessary and sufficient condition for an operator to be rank 1 in term of \( \partial \)-spectrum.
Lemma 3.2
Let \( R \in \mathcal {B}(X)\setminus \{0\}\), the following statements are equivalent.
-
1.
R has rank one.
-
2.
\( \Delta ^{*}(TRT \)) is a singleton for all operator \(T \in \mathcal {B}(X) \).
Proof
This is an immediate consequence of [23, Lemma 2.2]. \(\square \)
The next lemma will be used to show that if a surjective map \( \phi \) from \( \mathcal {B}(X) \) into itself preserves the \( \partial \)-spectrum of triple product of operators, then it’s automatically additive.
Lemma 3.3
Let \( T,S \in \mathcal {B}(X) \). Then, for every \(R \in \mathcal {F}_{1}(X)\), we have
Proof
Let \(R \in \mathcal {F}_{1}(X)\) such that \(R=x\otimes f\) where \(x \in X, ~ f\in X^{*} \). Then \( R(T+S)R=f((T+S)x)x\otimes f \), and
\(\square \)
Now, we are in a position to give the second main results in this paper.
Theorem 3.4
Let \(\phi :\mathcal {B}(X)\rightarrow \mathcal {B}(X)\) be a surjective map satisfying
Then there exists a scalar \( \alpha \in \mathbb {C} \) with \( \alpha ^{3}=1 \) and either there is a bounded invertible operator \( A: X \rightarrow X \) such that
or there is a bounded invertible operator \( C: X^{*} \rightarrow X \) such that
In the last case, X is automatically reflexive.
Proof
We break down the proof into several steps—
Step 1. \( \phi \) is injective and \( \phi (0)= 0\).
Let \(A,B \in \mathcal {B}(X)\) such that \( \phi (A) = \phi (B)\). Then for every \(R \in \mathcal {F}_{1}(X)\setminus \mathcal {N}_{1}(X)\), we have
By Lemma 3.1, we establish that \( A = B \) which proves that \( \phi \) is injective. Consequently \( \phi \) is a bijection and it is easy to show that \( \phi ^{-1} \) satisfies the Eq. 3.1.
Let us prove that \( \phi (0)= 0\). For every \(T \in \mathcal {B}(X)\), we have
By Lemma 3.1, we see that \( \phi (0)= 0\).
Step 2. \(\phi \) preserves rank one operators in both directions.
Let \(R \in \mathcal {F}_{1}(X)\), then \( Card(\Delta ^{*}(TRT) ) = 1 \) for all \(T \in \mathcal {B}(X)\). Which implies that \( Card(\Delta ^{*}(\phi (T)\phi (R)\phi (T) ) = 1 \) for all \(T \in \mathcal {B}(X)\). Its follows, by surjectivity of \( \phi \) and Lemma 3.2, that \( \phi (R)\in \mathcal {F}_{1}(X) \). Since \( \phi \) is a bijection and \( \phi ^{-1} \) satisfies (3.1), \( \phi \) preserves \( \mathcal {F}_{1}(X) \) in both directions.
Now, we show also that \(\phi \) preserves non-nilpotent rank one operators in both directions. Indeed, let \(R=x\otimes f\) where \(x \in X\) and \(f\in X^{*} \) with \( f(x)\ne 0\). Then \(\phi (R)=y\otimes g\) where \(y \in X\) and \( g\in X^{*}\). We have
Since \( f(x)\ne 0 \), then \( g(y) \ne 0 \). It follows that \( \phi (R) \in \mathcal {F}_{1}(X){\setminus }\mathcal {N}_{1}(X) \) and \( \phi \) preserves \( \mathcal {F}_{1}(X){\setminus } \mathcal {N}_{1}(X) \).
Step 3. \( \phi \) is a linear map.
Let \(T, S \in \mathcal {B}(X)\) and \( R \in \mathcal {F}_{1}(X){\setminus } \mathcal {N}_{1}(X) \). We have
Since \(R \in \mathcal {F}_{1}(X) \) is arbitrary and \( \phi \) is surjective, by Lemma 3.1, we get \( \phi (T+S) = \phi (T) + \phi (S) \). Let \( T \in \mathcal {B}(X) \) and \(\lambda \in \mathbb {C}\). For every \( R \in \mathcal {F}_{1}(X){\setminus } \mathcal {N}_{1}(X)\), we have
By Lemma 3.1 and the surjectivity of \( \phi \), we get \( \phi (\lambda T) = \lambda \phi (T)\), which proves that \( \phi \) is a linear map.
By linearity, we conclude that \( \phi \) preserves finite rank operators in both directions.
Step 4. \( \phi (I)=\alpha I\) for some scalar \(\alpha \in \mathbb {C}\) such that \(\alpha ^{3}=1 \).
Suppose that \( \phi (I) \) and I are linearly independent. Let \(x \in X\setminus \{0\}\) such that \( \phi (I)x \) and x are linearly independent and let \( f \in X^{*} \) such that \( f(x) = 1 \) and \( f(\phi (I)x) = 0 \). For \( R = x\otimes f \), step 2 and surjectivity of \( \phi \) ensures the existence of a non-nilpotent kank one operator \(T = y\otimes g \), where \( y \in \) X and \( g \in X^{*} \) with \( g(y) \ne 0 \) such that \( \phi (T) = R\). Therefore,
Thus \( g(y)=0 \). This contradiction shows that \( \phi (I)=\alpha I \) where \( \alpha \in \mathbb {C^{*}}\).
On the other hand, we have
This implies that \(\alpha ^{3}=1\), as desired.
Step 5. \( \phi \) has the desired form.
Let \( \varphi :=\alpha ^{-1}\phi \), and note that \(\varphi \) satisfies the Eq. (3.1) and \(\varphi (I)=I\). It follows, from Step 3 and Lemma 2.2, that \(\varphi \) takes either the form (2.4) or the form (2.5).
Suppose (2.4) holds. Let \( x\in X \) and \( f \in X^{*}\), we have
We get
By applying the closed graph theorem, we conclude that A is bounded. Moreover, \( Bf(Ax) = f(x) \) implies that \( A^{*}B=I_{X*} \). Thus, \( B=(A^{*})^{-1}=(A^{-1})^{*}\). So, for every \( x \in X \) and \( f \in X^{*}\), we have
Then
Now, let \( T \in \mathcal {B}(X )\) and \( R \in \mathcal {F}_{1}(X) \), we have
Since \( \varphi \) is surjective, Lemma 3.1 implies that
Consequently \( \phi (T) = \alpha ATA^{-1} \) with \( \alpha ^{3} = 1 \) for all \( T \in \mathcal {B}(X )\).
In a similar way, we will treat the case where \(\varphi \) takes the form (2.5). We get
Using the closed graph theorem, separately for C and D, we conclude that these operators are bounded. Therefore, both \( C^{*}: X^{*} \rightarrow X^{**} \) and \( D^{*}: X^{**} \rightarrow X^{*}\) are invertible.
Moreover, by a similar way as in the proof of Theorem 2.1, we show that X is reflexive and \( D = (C^{*})^{-1} =(C^{-1})^{*}\). Therefore,
Let T be an arbitrary operator in \( \mathcal {B}(X)\). For every \( R \in \mathcal {F}_{1}(X) \) we have
By Lemma 3.1 and surjectivity of \( \varphi \), we conclude that \(\varphi (T)=CT^{*}C^{-1} \) for all \(T \in \mathcal {B}(X)\). This implies that \(\phi (T)=\alpha CT^{*}C^{-1} \) for all \(T \in \mathcal {B}(X)\) with \( \alpha ^{3}=1 \), and finishes the proof. \(\square \)
References
Aupetit, B.: Spectrum-preserving linear mappings between Banach algebras or Jordan Banach algebras. J. Lond. Math. Soc. 62, 917–924 (2000)
Bourhim, A., Mashreghi, J.: Maps preserving the local spectrum of product of operators. Glasg. Math. J. 57(3), 709–718 (2015). https://doi.org/10.1017/S0017089514000585
Bourhim, A., Mashreghi, J.: Maps preserving the local spectrum of triple product of operators. Linear Multilinear Algebra. 63(4), 765–773 (2015)
Chan, J.T., Li, C.K., Sze, N.S.: Mapping preserving spectra of products of matrices. Proc. Am. Math. Soc. 135(4), 977–986 (2007)
Cui, J.-L., Hou, J.-C.: Additive maps on standard operator algebras preserving parts of the spectrum. J. Math. Anal. Appl. 282, 266–278 (2003)
Cui, J., Hou, J.: Linear maps between Banach algebras compressing certain spectral functions. Rocky Mountain J. Math. 34(2), 565–585 (2004)
Cui, J.L., Li, C.K.: Maps preserving peripheral spectrum of Jordan products of operators. Oper. Matr. 6, 129–146 (2012)
Cui, J., Li, C.K., Sze, N.S.: Unitary similarity invariant function preservers of skew products of operators. J. Math. Anal. Appl. 454, 716–729 (2017)
Hajighasemi, S., Hejazian, S.: Maps preserving some spectral domains of operator products. Linear Multilinear Algebra (2020). https://doi.org/10.1080/03081087.2020.1801567
Hou, J.C., Li, C.K., Wong, N.C.: Maps preserving the spectrum of generalized Jordan product of operators. Linear Algebra Appl. 432, 1049–1069 (2010)
Jafarian, A., Sourour, A.R.: Spectrum preserving Linear maps. J. Funct. Anal. 66, 255–261 (1986)
Kaplansky, I.: Algebraic and analytic aspects of operator algebras, Conference Board of the Mathematical Sciences Regional Conference Series in Mathematics, No. 1. Providence (RI): American Mathematical Society; 1970
Mbekhta, M.: Résolvant généralisé et théorie spectrale. J. Oper. Theory. 21, 69–105 (1989)
Mbekhta, M.: Linear Maps Preserving the Generalized Spectrum. Extracta Math. 22, 45–54 (2007)
Mbekhta, M.: Linear maps preserving the minimum and surjectivity modulus of operators. Oper. Matr. 4, 511–518 (2010)
Mbekhta, M., Ouahab, A.: Opérateur s-régulier dans un espace de Banach et théorie spectrale. Acta Sci. Math. (Szeged) 59, 525–543 (1994)
Miura, T., Honma, D.: A generalization of peripherally-multiplicative surjections between standard operator algebras. Cent. Eur. J. Math. 7(3), 479–486 (2009)
Molnár, L.: Some characterizations of the automorphisms of B(H) and C(X). Proc. Am. Math. Soc. 130, 111–120 (2005)
Muller, V.: Spectral theory of linear operators and spectral systems in banach algebras. Oper. Theory Adv. Appl. 139, 2 (2007)
Omladič, M., Šemrl, P.: Additive mappings preserving operators of rank one. Linear Algebra Appl. 182, 239–256 (1993)
Schmoeger, C.H.: Ein Spektralubhidungssatz. Arch. Math. 55, 484–489 (1990)
Sourour, A.R.: Invertibility preserving linear maps on L(X). Trans. Am. Math. Soc. 348, 13–30 (1996)
Zhang, W., Hou, J.: Maps preserving peripheral spectrum of Jordan semi-triple products of operators. Linear Algebra Appl. 435, 1326–1335 (2011)
Acknowledgements
The authors wish to express their thanks to the referee for carefully reading the paper and for giving valuable suggestions.
Funding
No funding was received for this manuscript.
Author information
Authors and Affiliations
Contributions
The authors have no competing interests to declare that are relevant to the content of this article.
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This file has been typeset with the option draft to illustrate that feature and its purpose.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Benbouziane, H., Daoudi, A., Kettani, M.EC.E. et al. Maps Preserving the \(\partial \)-Spectrum of Product or Triple Product of Operators. Mediterr. J. Math. 20, 312 (2023). https://doi.org/10.1007/s00009-023-02501-3
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00009-023-02501-3