Finite-codimensional compressions of symmetric and self-adjoint linear relations in Krein spaces (Q515964)

Let \(\mathcal G\) be a Krein subspace of a Krein space \(\mathcal K\) with finite codimension and let \(P\) the projection in \(\mathcal K\) onto \(\mathcal G\). Then a finite-codimensional compression \(A_0\) of an operator \(A\) in \(\mathcal K\) is defined as \[ A_0:=PA\left|_{\mathcal G \cap \text{dom}A} \right. \] and \(A_0\) is then considered as an operator in \(\mathcal G\). A similar definition is used in case of a linear relation \(A\). The following five questions are answered or partially answered. \begin{itemize} \item[1.] If \(A\) is a self-adjoint operator in the Krein space \(\mathcal K\) which has an invariant maximal nonnegative subspace, is the same true for \(A_0\)? \item[2.] If \(A\) is definitizable, is the same true for \(A_0\)? \item[3.] If a self-adjoint operator \(A\) in the Krein space \(\mathcal K\) is similar to a self-adjoint operator in a Hilbert space, is the same true for \(A_0\)? \item[4.] Let \(A\) and \(A_0\) be definitizable operators. If \(\infty\) is a critical point for one, is it also a critical point for the other? \item[5.] Let \(A\) be a closed densely defined symmetric operator in a Hilbert space. Do \(A\) and \(A_0\) have the same defect numbers? \end{itemize} The answer to the first question is partial in the sense that, under an additional condition, due to \textit{H.\ Langer} [Math.\ Ann.\ 146, 60--85 (1962; Zbl 0119.32003)], the answer is positive. The answer to the second question is yes for bounded operators, whereas in the unbounded case an additional assumption is needed. The answer to the third question is negative, which is shown by a simple matrix counterexample, whereas the answer to the fourth is (partially) positive under the additional assumption \(\mathcal G^\perp \subset \operatorname{dom}A\). Finally, the answer to the last question is positive.