Abstract
Let \({\mathcal {B}}(X)\) be the algebra of all bounded linear operators on an infinite dimensional complex Banach space X. For \(A\in {\mathcal {B}}(X)\), \(\sigma _{1}(A)\), \(\sigma _{2}(A)\), and \(\sigma _{3}(A)\) are the semi-Fredholm domain, the Fredholm domain, and the Weyl domain of A in the spectrum \(\sigma (A)\), respectively. For an integer \(k\ge 2\), let \((i_1,\ldots ,i_m)\) be a finite sequence with terms chosen from \(\{1,\ldots ,k\}\) and assume that at least one of the terms in \((i_1,\ldots ,i_m)\) appears exactly once. The generalized product of k operators \(A_1,\ldots ,A_k \in {\mathcal {B}}(X)\) is defined by
and includes the usual product and the triple product. In this paper we characterize the form of the map \(\phi \) from \({\mathcal {B}}(X)\) into itself which satisfies
for all \(A_1,\ldots ,A_k \in {\mathcal {B}}(X)\) and \(i \in \left\{ 1, 2, 3 \right\} \).
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Throughout this paper, X denotes an infinite-dimensional complex or real Banach space and \({\mathcal {B}}(X)\) the algebra of all bounded linear operators acting on X. The dual space of X will be denoted by \(X^*\) and the adjoint of an operator \( A\in {\mathcal {B}}(X)\) will be denoted by \(A^*\). For n a positive integer, \({\mathcal {F}}_{n}(X)\) denotes the set of operators of rank n on X and \({\mathcal {F}}(X)\) the set of all finite rank operators. For a non zero vector \(x\in X\) and a non zero linear functional f in the dual space \( X^{*}\), \( x\otimes f \) stands for the operator of rank one defined by \((x\otimes f)y:= f(y)x,\quad (y \in X )\), and every operator of rank at most one on \({\mathcal {B}}(X)\) can be written in this form. By \({\mathcal {N}}_{1}(X)\) we denote the set of rank one nilpotent operators on X. It is clear that \(x \otimes f \in {\mathcal {N}}_{1}(X)\) if and only if \(f(x)=0\). Let \( A\in {\mathcal {B}}(X)\), the spectrum, the null space, and the range of A are denoted by \(\sigma (A)\), N(A), and R(A) respectively. Recall that A is said to be lower semi-Fredholm if \(codimR(A)<\infty \), and A is said to be upper semi-Fredholm if \(dimN(A)<\infty \) and R(A) is closed. The semi-Fredholm operator is lower semi-Fredholm or a upper semi-Fredholm operator. A is said to be Fredholm if it is both lower and upper semi-Fredholm, in this case we define the index of A, indA, by
Obviously, any Fredholm operator has a finite index, if \(ind A = 0\) we say that A is Weyl operator. The semi-Fredholm domain, the Fredholm domain and the Weyl domain of the operator A are defined, respectively, by
It is clear that \(\rho _{w }(A) \subset \rho _{ f}(A) \subset \rho _{sf}(A)\). The semi-Fredholm domain, the Fredholm domain and the Weyl domain in the spectrum of the operator A are the subsets of the complex field defined, respectively, by
Obviously, for all \( A\in {\mathcal {B}}(X)\), \(\sigma _{1}(A)=\sigma (A)\cap \rho _{sf}(A)\), \(\sigma _{2}(A)=\sigma (A)\cap \rho _{f}(A)\) and \(\sigma _{3}(A)=\sigma (A)\cap \rho _{w}(A)\). Moreover it is clear that \(\sigma _{3 }(A) \subset \sigma _{ 2}(A) \subset \sigma _{1}(A)\).
For \(x\in X\) and \(f \in X^{*}\) if \(f(x)\ne 0 \), then
for all \(\lambda \in \mathbb {C}\) and \(i \in \{1,2,3\}\).
For a long time, there are many results concerning linear operators on a Banach space and others corresponding to inner-product spaces are realized(see, [3, 11]).
The study of maps on operator algebras preserving certain properties is a topic which attracts the attention of many authors (see, [1, 2, 4, 5, 7, 10, 13, 15]) and the references therein. In recent years a great activity has occurred in the question of relaxing the assumption of linearity or additivity (see, e.g. [1, 5, 10, 14, 15]). In [14], Shi. and Ji. proved that if X is a complex Banach space and if a surjective additive map \(\phi :{\mathcal {B}}(X) \rightarrow {\mathcal {B}}(X)\) satisfies \(\sigma _{i}(\phi (A))=\sigma _{i}(A)\) for all \(A \in {\mathcal {B}}(X)\) and \(i \in \left\{ 1, 2, 3 \right\} \), then one of the following forms is hold:
-
(i)
\(\phi (A)=TAT^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where T is an invertible linear operator in \({\mathcal {B}}(X)\).
-
(ii)
\(\phi (A)=TA^{*}T^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T: X^{*}\rightarrow X\) is bounded invertible linear operator. In this case, X must be reflexive.
In [9] Hajighasmi and Hejazian, replaced the additivity of \(\phi \) and the condition \(\sigma _{i}(\phi (A))=\sigma _{i}(A)\) for all \(A\in {\mathcal {B}}(X)\), and \(i \in \left\{ 1, 2, 3 \right\} \), by \(\sigma _{i}(\phi (A)\phi (B))=\sigma _{i}(AB)\) for all \(A, B\in {\mathcal {B}}(X)\) and \(i \in \left\{ 1, 2, 3 \right\} \). They showed that there exists a scalar \(\lambda \in \mathbb {C}\) with \(\lambda ^{2}=1\) such that either
-
(i)
\(\phi (A)=\lambda TAT^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T:X \rightarrow X\) is an invertible linear operator in \({\mathcal {B}}(X)\).
-
(ii)
\(\phi (A)=\lambda TA^{*}T^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T: X^{*} \rightarrow X\) is bounded invertible linear operator. In this case, X must be reflexive.
They showed also that if \(\phi \) is a surjective map from \( {\mathcal {B}}(X) \) into itself satisfying \(\sigma _{i}(\phi (A)\phi (B)\phi (A))=\sigma _{i}(ABA)\) for all \(A, B\in {\mathcal {B}}(X)\) and \(i \in \left\{ 1, 2, 3 \right\} \), then there exists \(\lambda \in \mathbb {C}\) with \(\lambda ^{3}=1\) such that either
-
(i)
\(\phi (A)=\lambda TAT^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T:X\rightarrow X\) is bijective linear operator in \({\mathcal {B}}(X)\).
-
(ii)
\(\phi (A)=\lambda TA^{*}T^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T: X^{*}\rightarrow X\) is bijective linear operator. In this case, X must be reflexive.
In [6], Bouramdane and Ech-cheérif El Kettani described maps from \({\mathcal {B}}({\mathcal {H}})\) into itself preserving some spectral domains in the spectrum of skew-products of operators, and established the following results.
Theorem 1.1
[6] Let \(\phi ~~:~~{\mathcal {B}}({\mathcal {H}}) \rightarrow {\mathcal {B}}({\mathcal {H}})\) be a map such that its range contains the operators of rank at most two. Then \(\phi \) satisfies
for all \(A,B \in {\mathcal {B}}({\mathcal {H}})\) and \(i \in \left\{ 1, 2, 3 \right\} \) if and only if there exist two unitary operators \(U,V \in {\mathcal {B}}({\mathcal {H}})\) such that
Theorem 1.2
[6] Let \(\phi ~~:~~{\mathcal {B}}({\mathcal {H}}) \rightarrow {\mathcal {B}}({\mathcal {H}})\) be a map such that its range contains the operators of rank at most two. Then \(\phi \) satisfies
for all \(A,B \in {\mathcal {B}}({\mathcal {H}})\) and \(i \in \left\{ 1, 2, 3 \right\} \) if and only if there exist a unitary operator \(U \in {\mathcal {B}}({\mathcal {H}})\) such that
Recall, for an integer \(k\ge 2\), let \((i_1,\ldots ,i_m)\) be a finite sequence with terms chosen from \(\{1,\ldots , k\}\), such that \(\{i_1,\ldots ,i_m\}=\{1,\ldots , k\}\) and at least one of the terms in \((i_1,\ldots ,i_m)\) appears exactly once. The generalized product of with m of k operators \(A_1,\ldots , A_k \in {\mathcal {B}}(X)\) is defined by
Evidently, The generalized product includes the usual product and the triple product.
Recently, there has been a lot of activities on describing maps that preserve the generalized product of operators; see for example [1, 2, 4, 5, 15] and the references therein. In this paper, we propose to determine the form of mappings that preserve some spectral domains in the spectrum of generalized product of operators.
The aim of this paper is to characterize maps \(\phi \) (with no additivity assumption ) from \({\mathcal {B}}(X)\) into itself which satisfies
for all \(A_1,\ldots ,A_k \in {\mathcal {B}}(X)\) and \(i \in \left\{ 1, 2, 3 \right\} \).
The main result is the following theorem.
Theorem 1.3
Let \(i \in \left\{ 1, 2, 3 \right\} \), and let \(\phi : {\mathcal {B}}(X) \rightarrow {\mathcal {B}}(X) \) be a surjective map which satisfies the following condition:
Then there exists \(\alpha \in \mathbb {C}\) with \(\alpha ^{k}=1\) such that either
-
(i)
\(\phi (A)=\alpha TAT^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T:X\rightarrow X\) is a bounded invertible linear (or conjugate linear) operator in \({\mathcal {B}}(X) \).
-
(ii)
\(\phi (A)=\alpha TA^{*}T^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T: X^{*}\rightarrow X\) is a bounded invertible linear (or conjugate linear) operator. In this case, X must be reflexive.
To establish our main results, we need the following well-known theorem. We recall that if \(h: \ \mathbb {C} \rightarrow \mathbb {C} \) is a homomorphism, then an additive map \(A:\ X \rightarrow X\) satisfying \(A(\alpha x)=h(\alpha )A(x) \ (x\in X,\alpha \in \mathbb {C})\) is called an h-quasilinear operator. If \(h(\alpha )={\bar{\alpha }}\), for \(\alpha \in \mathbb {C}\), then A is said to be a conjugate linear operator.
Lemma 1.4
[12, Theoreme 3.3] Let \(\phi : {\mathcal {F}}(X) \rightarrow {\mathcal {F}}(X) \) be a bijective additive map preserving rank one operators in both directions. Then there exists a ring automorphism \(h: \ \mathbb {C} \rightarrow \mathbb {C} \), and either there are h-quasilinear bijective mappings \(A:\ X \rightarrow X\) and \(B:\ X^{*} \rightarrow X^{*}\) such that:
or there are h-quasilinear bijective mappings \(C:\ X^{*} \rightarrow X\) and \(D:\ X \rightarrow X^{*}\) such that:
Note that, if in Lemma 1.4 the map \(\phi \) is linear, then h is the identity map on \(\mathbb {C}\) and so the maps A, B, C and D are linear.
2 Preliminaries
In this section, we collect and introduce some lemmas that will be used in the sequel. We begin with the following lemma that will be used frequently in what follows.
Lemma 2.1
For any \(x\in X \) and \(f\in X^{*}\), the following assertions hold:
-
(i)
\(\sigma _{i}(x\otimes f)= \{f(x)\} ~~~~~~ \text{ if } \text{ and } \text{ only } \text{ if }~~~~~~ f(x)\ne 0.\)
-
(ii)
\(\sigma _i(x\otimes f)= \emptyset ~~~~~~ \text{ if } \text{ and } \text{ only } \text{ if }~~~~~~ f(x)= 0.\)
Proof
[6, Lemme 2.3] \(\square \)
In the sequel, for an operator \(A \in {\mathcal {B}}(X)\) and \(i \in \{1,2,3\}\), we use a useful notation defined by Bouramdane and Ech-cheérif El Kettani [6] by
Let \(\phi \) be a map from \({\mathcal {B}}(X)\) into itself and r and s two nonegative integers such that \(r+s\ge 1\) and \(i \in \{1,2,3\}\). Then for all \( A, B \in {\mathcal {B}}(X)\), we have
We continue with the following identity principle.
Lemma 2.2
Let \(A, B\in {\mathcal {B}}(X)\), r and s two nonegative integers such that \(r+s\ge 1\) and \(i \in \{1,2,3\}\). Then the following statements are equivalent:
-
(i)
\(A=B\).
-
(ii)
\(\sigma _{i}(R^{r}AR^{s})=\sigma _{i}(R^{r}BR^{s})\) for all \(R \in {\mathcal {F}}_1(X) \backslash {\mathcal {N}}_1(X)\).
-
(iii)
\(\Lambda _{i}(R^{r}AR^{s})=\Lambda _{i}(R^{r}BR^{s})\) for all \(R \in {\mathcal {F}}_1(X)\).
Proof
The implication \(i)\Longrightarrow ii)\) is evident.
Let \(R=x\otimes f\) for \(x\in X \backslash \{0\}\) and \(f\in X^{*}\backslash \{0\}\) with \(f(x)\ne 0\), then we have \(R^{r}=f(x)^{r-1}x\otimes f\) and \(R^{s}=f(x)^{s-1}x\otimes f\). Consequently \(R^{r}AR^{s}=f(x)^{r+s-2}f(Ax)x\otimes f\) and \(R^{r}BR^{s}=f(x)^{r+s-2}f(Bx)x\otimes f\), thus
It follows that \(f(x)^{r+s-1}f(Ax)=f(x)^{r+s-1}f(Bx)\), so we have \(f(Ax)=f(Bx)\) for all \(x \in X \) and \(f \in X^*\), with \(f(x)\ne 0\).
Now, if \(f(x)=0\), let \(g \in X^{*}\) such that \(g(x)\ne 0\). Then by the first case
Then
Thus \(f(Ax)=f(Bx)\) in this case, it follows that \(f(Ax)=f(Bx)\) for all \(x \in X\) and \(f \in X^* \). Thus \(Ax=Bx\) for all \(x \in X\). Which proves that \(A=B\). \(\square \)
The following lemma is a characterization of rank one operators in term of spectral domains in the spectrum.
Lemma 2.3
Let \(A \in {\mathcal {B}}(X)\), s and r are two nonegative integers, and \(i \in \{1,2,3\}\). The following statements are equivalent
-
(i)
\(A \in {\mathcal {F}}_{1}(X)\).
-
(ii)
\(\sigma _{i}(R^{r}AR^{s})\) contains at most one element for all \(R \in {\mathcal {F}}_2(X)\).
-
(iii)
\(\sigma _{i}(R^{r}AR^{s})\) contains at most one element for all \(R \in {\mathcal {B}}(X)\).
Proof
[8, Lemme 2.4] \(\square \)
We end this section with the following lemma which is a useful observation which allow us to prove that a map satisfying the equation (3.1) is additive.
Lemma 2.4
Let \(R \in {\mathcal {F}}_{1}(X)\) and \(A, B \in {\mathcal {B}}(X)\), and \(i \in \{1,2,3\}\). Then, The following statements hold:
-
(i)
\(\Lambda _{i}((A+B)R)=\Lambda _{i}(AR)+\Lambda _{i}(BR)\).
-
(ii)
\(\Lambda _{i}(R(A+B)R)=\Lambda _{i}(RAR)+\Lambda _{i}(RBR)\).
Proof
[6, Lemme 2.4] \(\square \)
3 Main Result
Since all the necessary ingredients are collected in the preliminary section we will state and prove the promised main result. Let \(A, B \in {\mathcal {B}}(X)\), set \(A_{i_{p}}=B\) and \(A_{i_{j}}=A\) for \(j\ne p\) where \(i_{p}\) is the term which appears exactly once in \((i_1,\ldots ,i_m)\). Note that \(A_1*A_2*\cdots *A_k=A^rBA^s\) for some nonegative integers r and s such that \(r+s=m-1\). Then, the Theorem 1.3 is a consequence of the following one.
Theorem 3.1
Let r and s be two nonegative integers with \(r+s \ge 1\), and \(i \in \{1,2,3\}\). Let \(\phi : {\mathcal {B}}(X) \rightarrow {\mathcal {B}}(X) \) be a surjective map which satisfies the following condition:
Then there exists \(\alpha \in \mathbb {C}\) with \(\alpha ^{r+s+1}=1\) such that either
-
(i)
\(\phi (T)=\alpha ATA^{-1}\) for all \(T \in {\mathcal {B}}(X) \), where \(A:X\rightarrow X\) is a bounded invertible linear (or conjugate linear) operator in \({\mathcal {B}}(X) \).
-
(ii)
\(\phi (T)=\alpha CT^{*}C^{-1}\) for all \(T \in {\mathcal {B}}(X) \), where \(C: X^{*}\rightarrow X\) is a bounded invertible linear (or conjugate linear) operator. In this case, X must be reflexive.
Proof
Assume that \(\phi \) is a surjective map from \({\mathcal {B}}(X)\) into itself which satisfies the Eq. (3.1). We divide the proof into several steps.
Step 1. \(\phi \) is injective.
Let \(A,B \in {\mathcal {B}}(X)\) such that \(\phi (A)=\phi (B)\). Then, by hypothesis, for every \(T\in {\mathcal {F}}_{1}(X) \backslash {\mathcal {N}}_{1}(X)\) we have
It follows, by Lemma 2.2 that \(A=B\), and so \(\phi \) is injective. Thus \(\phi \) is bijective since it is assumed to be surjective, moreover \(\phi ^{-1}\) satisfies the equation (3.1).
Step 2. \(\phi \) preserves \({\mathcal {F}}_{1}(X)\) and \({\mathcal {N}}_{1}(X)\) in both directions.
Let R be a rank one operator, and \(A \in {\mathcal {B}}(X) \) then, by Lemma 2.3, \(\sigma _{i}(A^{r}RA^{s})\) contains at most one element. It follows by hypothesis that \(\sigma _{i}(\phi (A)^{r}\phi (R)\phi (A)^{s})\) contains at most one element for every \(A\in {\mathcal {B}}(X)\), again by the surjectivity of \(\phi \) we deduce that \(\phi (R)\) is a rank one operator. Since \(\phi \) is bijective and \(\phi ^{-1}\) satisfies the equation (3.1), we conclude that \(\phi \) preserves rank one operators in both directions.
Now, let \(R \in {\mathcal {F}}_1(X)\), by hypothesis, we have \(\sigma _{i}(R^{r+s+1})=\sigma _{i}(\phi (R)^{r+s+1})\). If \(R \in {\mathcal {N}}_1(X)\), then \(R^{r+s+1} \in {\mathcal {N}}_1(X)\) and so \(\sigma _{i}(\phi (R)^{r+s+1})=\emptyset \). It follows that \(\phi (R) \in {\mathcal {N}}_1(X)\). The inverse is similar.
Step 3. \(\phi \) is linear and preserves \({\mathcal {F}}(X)\) in both directions.
Let us show first that \(\phi \) is homogeneous. To do that consider \(A \in {\mathcal {B}}(X)\) and \(\lambda \in \mathbb {C}\). Note that \(\sigma _{i}(\lambda R)=\lambda \sigma _{i}(R)\) for every \(R \in {\mathcal {F}}_{1}(X) \backslash {\mathcal {N}}_{1}(X)\), by the Eq. (1.1). Thus for every \(R \in {\mathcal {F}}_{1}(X) \backslash {\mathcal {N}}_{1}(X)\), we have
And by hypothesis
This implies that
By appliying Lemma 2.2 and Step 2., we conclude that
It remains to show that \(\phi \) is additive. To do that let \(A, B \in {\mathcal {B}}(X)\) and \(R \in {\mathcal {F}}_{1}(X)\), then by appliying the Lemma 2.4, we have
So by (2.1)
And by appliying the Lemma 2.2 and Step 2., we get \(\phi (A+B)=\phi (A)+\phi (B)\quad \text{ for } \text{ all }\quad A, B \in {\mathcal {B}}(X)\), and \(\phi \) is additive.
Finally, since each \(T \in {\mathcal {F}}(X)\) is a finite linear combination of rank one operators and since \(\phi \) preserves \({\mathcal {R}}_{1}(X)\) in both directions, we have the result by linearity.
Step 4. \(\phi \) takes the desired form.
It follows by has been already proved in the Step 3. and Lemma 1.4 that \(\phi \) takes one of the following forms
-
(1)
There exists a bounded invertible linear (or conjugate linear) operator \(A:X\rightarrow X\) and \(B:X^{*}\rightarrow X^{*}\) such that
$$\begin{aligned} \phi (x\otimes f)=Ax\otimes Bf, ~~~~ \text{ for } \text{ all }~~~~ x\in X ~~~~\text{ and }~~~~ f \in X^{*}. \end{aligned}$$ -
(2)
There exists a bounded invertible linear (or conjugate linear) \(C:X^*\rightarrow X\) and \(D:X\rightarrow X^*\) such that
$$\begin{aligned} \phi (x\otimes f)=Cf\otimes Dx, ~~~~ \text{ for } \text{ all }~~~~ x\in X~~~~ \text{ and }~~~~ f \in X^{*}. \end{aligned}$$
Assume that \(\phi \) has the first form. Let us show that \(\phi (I)=\alpha I\) with \( \alpha ^{r+s+1}=1\).
Since \(\phi \) is surjective, there exist \(S\in {\mathcal {B}}(X)\) such that \(I=\phi (S)\), then \(S=\lambda I\). Otherwise assume that there exist \(x\in X\) such that Sx and x are linearly independent. Then there is \(f\in X^{*}\) such that \(f(Sx)=0\) and \(f(x)=1\). Note that \(Bf(Ax)\ne 0\). Then by Lemma 2.1 we have
wich is a contradiction. Since \(\phi (\lambda I)=I\), and by linearity of \(\phi \), we conclude that \(\phi (I)=\alpha I\).
Let \(f \in X^{*}\) and \(x \in X\) such that \(f(x)=1\), then by Lemma 2.1 we obtain
We conclude that
On the other hand, we have
Thus
Otherwise
So
By combining of (3.2), (3.3) and (3.4), we get \( \alpha ^{r+s+1}=1\).
Now, let us show that \(\phi (x\otimes f)=\alpha Ax\otimes fA^{-1}\) for every \(x\in X\) and \(f\in X^{*}\). Let \(x\in X\) and \(f\in X^{*}\), we have
Then
Hence
Let us prove that A is continuous and \(B=\alpha (A^{*})^{-1}\). To do that consider \((x_{n})_{n \ge 0}\) a sequence of vectors of X such that \(\lim _{n \rightarrow +\infty }x_{n}= x \in X \) and \( \lim _{n \rightarrow +\infty }Ax_{n}=y \in X \). We have
Since B is bijective, we see that \(Ax=y\) and the closed graph theorem tells us that A is continuous. It follows from (3.5) that \(\alpha f(x)=A^{*}Bf(x)\), for all \(x\in X\) and \(f\in X^{*}\), wich implies that \(A^{*}B=\alpha I_{X^{*}}\); as desired. Thus, for all \(x\in X\) and \(f\in X^{*}\), we have \(\phi (x\otimes f)=\alpha Ax\otimes fA^{-1}\).
Now, let \(T\in {\mathcal {B}}(X) \) and \(R \in {\mathcal {F}}_{1}(X) \backslash {\mathcal {N}}_{1}(X)\), we have by hypothesis
This implies, by Lemma 2.2, that \(T=\alpha ^{r+s}A^{-1}\phi (T)A\), so \(\phi (T)=\alpha ATA^{-1}\).
Now suppose that \(\phi \) has the second form. Since \(\phi \) is surjective, so there exist \(S\in {\mathcal {B}}(X)\) such that \(I=\phi (S)\). Then \(S=\lambda I\). Otherwise assume there exist \(x\in X\) such that Sx and x are linearly independent. Then there exist a linear form \(f\in X^{*}\) such that \(f(Sx)=0\) and \(f(x)=1\). Note that \(Dx(Cf)\ne 0\). Then by Lemma 2.1
wich is a contradiction. Hence \(\phi (\lambda I)=I\), and by linearity of \(\phi \), we conclude that \(\phi (I)=\alpha I\).
Let \(f \in X^{*}\) and \(x \in X\) such that \(f(x)=1\), then by Lemma 2.1 we have
We conclude that
and,
Hence
Otherwise
Hence
By combining (3.6), (3.7) and (3.8), we conclude that \( \alpha ^{r+s+1}=1\).
Now, choose \(f \in X^{*}\), we have
Hence \(Dx(Cf)=\dfrac{1}{\alpha ^{r+s}}f(x)\). Since \(\alpha ^{r+s+1}=1\), thus
Consequently, by similar reasoning as the first case and by using the closed graph theorem, separately for each of C and D, we conclude that these operators are bounded. Therfore, both \(C^{*}\,\ X^{*}\ \rightarrow \ X^{**} \) and \(D^{*}\,\ X^{**}\ \rightarrow \ X^{*} \) are invertible. Moreover, if \(\beta \) is the canonical embedding of X in \(X^{**}\). Then by (3.9) we have \(D^{*}\circ \beta \circ C=\alpha I_{X^{*}}\) which implies that \(\beta \circ C=\alpha (D^{*})^{-1}\). Since both C and \((D^{*})^{-1}\) are surjective, \(\beta \) is also surjective and hence X is reflexive. Identifying X with \(X^{**}\), so if \(f(x)=1\) then \(D^{*}C=\alpha I_{X^{*}}\). Similarly, it follows from (3.9) that \(C^{*}D=\alpha I\) and so \(D=\alpha (C^{*})^{-1}=\alpha (C^{-1})^{*}\). So \(\phi (R)=\alpha CR^{*}C^{-1}\) for all \(R \in {\mathcal {F}}_{1}(X)\). Since the spectrum of a bounded linear operator coincides with the spectrum of its adjoint, for arbitrary \(T \in {\mathcal {B}}(X)\), we obtain that
For all \(R \in {\mathcal {F}}_{1}(X)\backslash {\mathcal {N}}_{1}(X)\). By Lemma 2.2 and step 2., we get \(\phi (T)=\alpha CT^{*}C^{-1}\).
By taking \(r=0\) and \(s=1\) (resp. \(r=s=1\))the [9, Theoreme 2.3](resp. [9, Theoreme 3.5]) becomes corollary of Theoreme 3.1.
Corollary 3.2
Let \(i \in \left\{ 1,2,3\right\} \) and let \(\phi : {\mathcal {B}}(X)\ \rightarrow \ {\mathcal {B}}(X)\) be a surjective map satisfying
Then there exists a scalar \(\alpha \) with \(\alpha ^{2}=1\) such that either
-
(1)
\(\phi (A)=\alpha TAT^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T:X\rightarrow X\) is bijective linear operator in \({\mathcal {B}}(X) \).
-
(2)
\(\phi (A)=\alpha TA^{*}T^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T: X^{*}\rightarrow X\) is bijective linear operator. In this case, X must be reflexive.
Corollary 3.3
Let \(i \in \left\{ 1,2,3\right\} \) and let \(\phi : {\mathcal {B}}(X)\ \rightarrow \ {\mathcal {B}}(X)\) be a surjective map satisfying
Then there exists a scalar \(\alpha \) with \(\alpha ^{3}=1\) such that either
-
(1)
\(\phi (A)=\alpha TAT^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T:X\rightarrow X\) is bijective linear operator in \({\mathcal {B}}(X) \).
-
(2)
\(\phi (A)=\alpha TA^{*}T^{-1}\) for all \(A \in {\mathcal {B}}(X) \), where \(T: X^{*}\rightarrow X\) is bijective linear operator. In this case, X must be reflexive.
\(\square \)
Data Availibility
Data sharing not applicable to this article as no datasets were generated or analysed during the current study.
References
Abdelali, Z., Achchi, A., Marzouki, R.: Maps preserving the local spectral radius zero of generalized product of operators. Linear Multilinear Algebra 67(10), 2021–2029 (2019)
Achchi, A.: Maps preserving the inner local spectral radius zero of generalized product of operators. Rend. Circ. Mat. Palermo II. Ser 2 68(2), 355–362 (2019)
Alexandra, M., Bianca, M.: Ulam stability in real inner-product spaces. Constr. Math. Anal. 3(3), 113–115 (2020)
Benbouziane, H., Bouramdane, Y., Ech-Cherif El Kettani, M.: Maps preserving local spectral subspaces of generalised product of operators. Rend. Circ. Mat. Palermo II. Ser 2, 1–10 (2019)
Bouramdane, Y., Ech-Cherif El Kettani, M., Elhiri, A., Lahssaini, A.: Maps preserving fixed points of generalized product of operators. Proyecciones J. Math. https://doi.org/10.22199/issn.0717-6279-2020-05-0071
Bouramdane, Y., Ech-Cherif El Kettani, M.: Maps preserving some spectral domains of skew products of operators. Linear Multilinear Algebra. https://doi.org/10.1080/03081087.2021.1976716 (2021)
Dolinar, G., Du, S., Hou, J., Li, C.K., Legi\(\breve{s}\)a, P.: General preservers of invariant subspace lattices. Linear Algebra Appl. 429, 100–109 (2008)
Hou, J., Li, C.K., Wong, N.C.: Jordan isomorphisms and maps preserving spectra of certain operator products. Studia Math. 184(1), 31–47 (2008)
Hajighasemi, S., Hejazian, S.: Maps preserving some spectral domains of operator products. Linear Multilinear Algebra. https://doi.org/10.1080/03081087.2020.1801567
Jaafarian, A., Sourour, A.: Linear maps that preserve the commutant, double commutant or the lattice of invariant subspaces. Linear Multilinear Algebra 38, 117–129 (1994)
Michael, Gil: On matching distance between eigenvalues of unbounded operators. Constr. Math. Anal. 5(1), 46–53 (2022)
Omladic, M., Semrl, P.: Additive mappings preserving operators of rank one. Linear Algebra Appl. 182, 239–256 (1993)
Petek, T.: Mappings preserving the idempotency of products of operators. Linear Multilinear Algebra 58(7), 903–925 (2010)
Shi, W., Ji, G.: Additive maps preserving the semi-Fredholm domain in spectrum. Ann. Funct. Anal. 7(2), 254–260 (2016)
Zhang, W., Hou, J.-C.: Maps preserving peripheral spectrum of generalized products of operators. Linear Algebra Appl. 468, 87–106 (2015)
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sector.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Communicated by Fabrizio Colombo
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This article is part of the topical collection “Spectral Theory and Operators in Mathematical Physics” edited by Jussi Behrndt, Fabrizio Colombo and Sergey Naboko.
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
Ech-Cherif El Kettani, M., Siar, E. On the Some Spectral Domains in the Spectrum Preserves. Complex Anal. Oper. Theory 17, 100 (2023). https://doi.org/10.1007/s11785-023-01364-6
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11785-023-01364-6