Additive results for the Wg-Drazin inverse (Q967502): Difference between revisions

Let \(X\) and \(Y\) be Banach spaces and \(\mathcal B (X,Y)\) denote the set of all bounded linear operators from \(X\) into \(Y\). Set \(\mathcal B (X)=\mathcal B (X,X)\). For a fixed nonzero \(W \in \mathcal B (Y,X)\), let \(\mathcal B _W(X,Y)\) be the Banach algebra \(\mathcal B (X,Y)\) endowed with the multiplication \(*\) defined by \(A*B=AWB\) and the norm \({\| A \|}_W = \| A \| \| W \| \). An operator \(A \in \mathcal B (X,Y)\) is called Wg-Drazin invertible if \(A\) is quasipolar in \(\mathcal B _W(X,Y)\) (meaning that there exists \(C \in \mathcal B _W(X,Y)\) such that \(A*C=C*A\), \(C*A*C=C\) and \(A-A*C*A\) is quasinilpotent). When it exists, the Wg-Drazin inverse of \(A\) is denoted by \(A^{d,W}\). In the main result of the paper under review, the authors present sufficient conditions under which the Wg-Drazin inverse of the sum \(A+B\) of the two Wg-Drazin invertible operators \(A\) and \(B\) is Wg-Drazin invertible. In such a case, an explicit formula (involving infinite series representations in terms of \(A,B\) and \(W\)) is also presented. We only present that part of the main result (Theorem 2.2) which guarantees that the sum is Wg-Drazin invertible, the explicit formula being too complicated to be included here: For a nonzero \(W \in \mathcal B (Y,X)\), let \(A,B \in \mathcal B (X,Y)\) both be Wg-Drazin invertible. If \(A^{d,W}WB=0, AWB^{d,W}=0\) and \((I-WBWB^{d,W})WAWB(I-WAWA^{d,W})=0\), then \(A+B\) is Wg-Drazin invertible. A number of consequences are then presented as corollaries of the main result. The reviewer finds it curious that Theorem 2.1, purportedly a particular case of the main result, is not derived as a corollary (of the main result), but is proved earlier.