Nonlinear maps preserving the inner local spectral radius (Q2517007)

Let \(X\) and \(Y\) be infinite dimensional complex Banach spaces and \(\mathcal{B}(X)\) the algebra of all bounded linear operators on \(X\). For \(T\in \mathcal{B}(X)\) and \(x\in X\), \(\iota _{T}(x)\) will denote the inner local spectral radius of \(T\) at \(x\). In the paper under review, the author first describes the form of all surjective maps \(\varphi :\mathcal{B}(X)\rightarrow \mathcal{B}(X)\) that satisfy \[ \iota _{T-S}(x)=0\quad \text{if and only if\quad }\iota _{\varphi (T)-\varphi (S)}(x)=0 \] for every \(x\in X\) and \(T,S\in \mathcal{B}(X)\). The author proves that \(\varphi \) is such a map if and only if there exists a nonzero scalar \(c\in \mathbb{C}\) and an operator \(R\in \mathcal{B}(X)\) such that \(\varphi (T)=cT+R\) for all \(T\in \mathcal{B}(X)\). With a very similar proof, the author also characterizes surjective maps on \(\mathcal{B}(X)\) that satisfy \(\iota _{T+S}(x)=0\) if and only if \(\iota _{\varphi (T)+\varphi (S)}(x)=0\) for every \(x\in X\) and \(T,S\in \mathcal{B}(X)\). Let \(x_{0}\in X\) and \(y_{0}\in Y\) be nonzero fixed vectors. In the second part of the paper, the author characterizes bicontinuous bijective maps \(\varphi :\mathcal{B}(X)\rightarrow \mathcal{B}(Y)\) for which there are constants \(m,M>0\) such that \[ m\cdot\iota _{T\pm S}(x_{0})\leq \iota _{\varphi (T)\pm \varphi (S)}(y_{0})\leq M\cdot\iota _{T\pm S}(x_{0}),\quad T,S\in \mathcal{B}(X). \leqno{(\ast)} \] The paper has five sections. In the second section, the author provides some preliminaries, e.g., Lemma 2.2 from \textit{J. Bračič} and \textit{V. Müller} [Stud. Math. 194, No. 2, 155--162 (2009; Zbl 1182.47004)] which is the backbone of the proofs of the main results. The main results are stated and proved in the third and fourth sections. In the last section, the author provides some concluding remarks and poses as open question whether the form of all bijective maps \(\varphi \) from \(\mathcal{B}(X)\) onto \(\mathcal{B}(Y)\) that satisfy \((\ast)\) remains the same when the bicontinuity assumption is omitted.