On sectorial matrices (Q1405030)

Let \(x,y\) be vectors in the complex space \({\mathbb C}^n\). Let \(A\) be an \(n\times n\) matrix with entries in \({\mathbb C}\). Then \((Ax,x)=x^\star Ax\). The numerical range of the operator \(A\) is the set \(W(A)=\{(Ax,x)\mid \|x\|=1\}\). A linear operator is called sectorial with the vertex at the origin and a semiangle \(\alpha\in [0,{{\pi}\over{2}})\) if \(W(A)\subseteq\Theta(\alpha)=\{z \mid |\text{arg} z|\leq\alpha; z\in{\mathbb C}\}\). Define \(A_R={{A+A^\star}\over{2}}\) and \(A_I={{A-A^\star}\over{2}}\). The authors prove two criteria for a square matrix to be sectorial: 1) \(A_R\) is positive semidefinite and \(\text{rank} A\leq\text{rank} A_R\); 2) all principal minors of \(A_R\) are nonnegative and the corresponding principal minors of \(A\) and \(A_R\) vanish simultaneously. The authors obtain much more information on sectorial matrices. Among other things, this involves normal forms, the pseudo-inverse of \(A_R\), and, in the case of a block matrix, the Schur complement.