Pathological hypercyclic operators (Q2490663)

A bounded or unbounded linear operator \(T\) acting on a complex Banach space \(X\) is said to be hypercyclic with a hypercyclic vector \(x\in X\) if the orbit \(\{T^nx:n\geq0\}\) is norm dense in \(X\). If \(T\) is ``bounded'', then the following statements hold. (1) Every hypercyclic vector for \(T\) is a hypercyclic vector for all positive powers \(T^n\) of \(T\); see [\textit{S.~I.\ Ansari}, J.\ Funct.\ Anal.\ 128, No.~2, 374--383 (1995; Zbl 0853.47013)]. (2) For every \(\lambda\in{\mathbb C}\) with \(| \lambda| =1\), the operators \(T\) and \(\lambda T\) have the same sets of hypercyclic vectors; see [\textit{F.~León--Saavedra} and \textit{V.~Müller}, Integral Equations Oper.\ Theory 50, No.~3, 385--391 (2004; Zbl 1079.47013)]. Yet it is still unknown whether or not \(T\oplus T\) is hypercylic whenever \(T\) is. This problem was raised by \textit{D.~Herrero } [J.\ Oper.\ Theory 28, No.~1, 93--103 (1992; Zbl 0806.47020)]. In the paper under review, the author gives examples of ``unbounded'' hypercyclic operators \(T\) defined on the whole of \(X\) for which \(T^2\), \(-T\) and \(T\oplus T\) are not hypercyclic. He also provides an example of an unbounded Hilbert space operator for which all nonzero vectors are hypercyclic. This is equivalent to saying that the operator he constructs has no nontrivial invariant closed subsets. A bounded Banach space operator of this kind was constructed by \textit{C.~J.\ Read} [Isr.\ J.\ Math.\ 63, No.~1, 1--40 (1988; Zbl 0782.47002)]. It is still not known whether there is a bounded linear operator on a Hilbert space which has no nontrivial closed invariant subspaces (or subsets).