When is Hyperlat\(T\) = Hyperlat\(f(T)\) in finite dimension? (Q1855416)

Let \(T\) be an operator on \(\mathbb{C}^n\) and let \(f\) be an analytic function in a neighbourhood of \(\sigma(T)\) (the spectrum of \(T\)). Let \(\operatorname{Hyperlat}T\) denote the hyperinvariant subspace lattice of \(T\). The main result of the paper gives the following necessary and sufficient condition under which \(\operatorname{Hyperlat}T =\operatorname{Hyperlat}f(T)\): (a) \(f\) is one-to-one on \(\sigma(T)\) and, if there is \(\lambda\in\sigma(T)\) such that \(f'(\lambda)=0\), then either (b1) \(f''(\lambda)=0\) and all Jordan blocks corresponding to \(\lambda\) have dimension \(1\), or (b2) \(f''(\lambda)\neq 0\) and all Jordan blocks corresponding to \(\lambda\) have the same odd dimension.