Quasi-diagonalization of Hankel operators - Archive ouverte HAL Accéder directement au contenu
Article Dans Une Revue Journal d'analyse mathématique Année : 2017

Quasi-diagonalization of Hankel operators

Résumé

We show that all Hankel operators $H$ realized as integral operators with kernels $h(t+s)$ in $L^2 ({\Bbb R}_{+}) $ can be quasi-diagonalized as $H= {\sf L}^* \Sigma {\sf L} $. Here ${\sf L}$ is the Laplace transform, $\Sigma$ is the operator of multiplication by a function (distribution) $\sigma(\lambda)$, $\lambda\in {\Bbb R}$. We find a scale of spaces of test functions where ${\sf L} $ acts as an isomorphism. Then ${\sf L}^*$ is an isomorphism of the corresponding spaces of distributions. We show that $h= {\sf L}^* \sigma$ which yields a one-to-one correspondence between kernels $h(t)$ and sigma-functions $\sigma(\lambda)$ of Hankel operators. The sigma-function of a self-adjoint Hankel operator $H$ contains substantial information about its spectral properties. Thus we show that the operators $H$ and $\Sigma$ have the same numbers of positive and negatives eigenvalues. In particular, we find necessary and sufficient conditions for sign-definiteness of Hankel operators. These results are illustrated at examples of quasi-Carleman operators generalizing the classical Carleman operator with kernel $h(t)=t^{-1}$ in various directions. The concept of the sigma-function directly leads to a criterion (equivalent of course to the classical Nehari theorem) for boundedness of Hankel operators. Our construction also shows that every Hankel operator is unitarily equivalent by the Mellin transform to a pseudo-differential operator with amplitude which is a product of functions of one variable only (of $x\in{\Bbb R}$ and of its dual variable).

Dates et versions

hal-01342755 , version 1 (06-07-2016)

Identifiants

Citer

Dimitri R Yafaev. Quasi-diagonalization of Hankel operators. Journal d'analyse mathématique, 2017, 133 (1), pp.133-182. ⟨10.1007/s11854-017-0030-7⟩. ⟨hal-01342755⟩
140 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More