Relative \(\sigma\)-boundedness of linear operators (Q1390451): Difference between revisions

Let \(L\) and \(M\) be linear operators acting between Banach spaces; \(L\) continuous and everywhere defined, \(M\) closed and densely defined. In the first part of the paper the authors give the Laurent expansion of \(\mu \rightarrow (\mu L-M)^{-1}\) at infinity if \(M\) is spectrally bounded with respect to \(L\), that is \(\{\mu\in\mathbb C: \mu L-M\;\text{is not injective or not surjective}\}\) is bounded. They use a decomposition of \(L\) and \(M\) by the ``spectral'' projections \[ \frac{1}{2\pi i} \int_{| \mu| =r} (\mu L-M)^{-1}L d\mu \quad \text{and}\quad \frac{1}{2\pi i} \int_{| \mu| =r} L(\mu L-M)^{-1} d\mu , \] \(r\) sufficiently large; and they characterize the situation when infinity is not an essential singularity of \(\mu \rightarrow (\mu L-M)^{-1}\). In the second part the authors prove that a decomposition of \(L\) and \(M\) with corresponding properties is sufficient for an operator \(M\) to be spectrally bounded with respect to \(L\) and infinity is not an essential singularity of \(\mu \rightarrow (\mu L-M)^{-1}\). (The English translation is quite inadequate).