Matrix Completions, Norms, and Hadamard Products (Q4039288)

The author obtains a necessary and sufficient condition for \(X_ 0\in S\subset H_ n\) to attain the maximum in the problem \(\max\{\lambda_{\min}(A+X):X\in S\}\), where \(A\in H_ n\) is a fixed matrix, \(H_ n\) is the space of \(n\times n\) Hermitian matrices, and \(S\) is a closed convex set. Using this condition the author gives an elementary and unified proof of the following results. 1. For \(A\in M_{n,n}\), \(\omega(A)=\max\{x^*Ax:x^*x=1\}\leq 1\) if and only if there is a matrix \(Z\in H_ n\) such that \(\left(\begin{matrix} I+Z & A \\ A & I-Z\end{matrix}\right)\geq 0\). 2. For any \(A\in M_{m,n}\) \(\max\bigl\{\| A\circ B\|_ \infty:\| B\|_ \infty\leq 1\bigr\}\leq 1\) if and only if there are matrices \(P\in H_ m\), \(Q\in H_ n\) such that \(\left(\begin{matrix} P & A \\ A & Q \end{matrix}\right)\geq 0\), \(P\circ I\leq I\), \(Q\circ I\leq I\). 3. For \(A\in M_{n,n}\) \(\max\{\omega(A B):\omega(B)\leq 1\}\leq 1\) if and only if there is a matrix \(P\in H_ n\) such that \(\left(\begin{matrix} P & A \\ A & P \end{matrix}\right)\geq 0\), \(P\circ I\leq I\).