Finite rank perturbations of normal operators: hyperinvariant subspaces and a problem of Pearcy

TL;DR

This paper investigates the existence of invariant and hyperinvariant subspaces for finite rank perturbations of normal operators, providing new conditions under which such subspaces exist, thus advancing the understanding of operator perturbations.

Contribution

It introduces new criteria for the existence of invariant subspaces in finite rank perturbations of normal operators, improving previous results and addressing a long-standing open problem.

Findings

Invariant subspaces exist under specific Fourier coefficient conditions.

Operators have hyperinvariant subspaces unless they are scalar multiples of the identity.

Results extend to finite rank perturbations of diagonalizable normal operators.

Abstract

Finite rank perturbations of diagonalizable normal operators acting boundedly on infinite dimensional, separable, complex Hilbert spaces are considered from the standpoint of view of the existence of invariant subspaces. In particular, if $T=D_\Lambda+u\otimes v$ is a rank-one perturbation of a diagonalizable normal operator $D_\Lambda$ with respect to a basis $\mathcal{E}=\{e_n\}_{n\geq 1}$ and the vectors $u$ and $v$ have Fourier coefficients $\{\alpha_n\}_{n\geq 1}$ and $\{\beta_n\}_{n\geq 1}$ with respect to $\mathcal{E}$ respectively, it is shown that $T$ has non trivial closed invariant subspaces provided that either $u$ or $v$ have a Fourier coefficient which is zero or $u$ and $v$ have non zero Fourier coefficients and $$ \sum_{n\geq 1} |\alpha_n|^2 \log \frac{1}{|\alpha_n|} + |\beta_n|^2 \log \frac{1}{|\beta_n|} < \infty.$$ As a consequence, if $(p,q)\in (0,2]\times (0,2]$ are such $\sum_{n\geq 1} (|\alpha_n|^p + |\beta_n|^q )< \infty,$ it is shown the existence of non trivial closed invariant subspaces of $T$ whenever $$(p,q)\in (0,2]\times (0,2]\setminus \{(2, r), (r, 2):\; r\in(1,2]\}.$$ Moreover, such operators $T$ have non trivial closed hyperinvariant subspaces whenever they are not a scalar multiple of the identity. Likewise, analogous results hold for finite rank perturbations of $D_\Lambda$. This improves considerably previous theorems of Foia\c{s}, Jung, Ko and Pearcy, Fang and Xia and the authors on an open question explicitly posed by Pearcy in the seventies.