P-matrices: an equivalent characterization (Q1097330)

A square matrix is a P-matrix if all its principal minors are positive. The class of P-matrices includes the positive definite matrices as well as totally positive matrices. P-matrices have interesting properties relating to stability so that different characterizations of such matrices seem to be of importance. Let M be any subset of \(N=\{1,2,...,n\}\) and let \(I_ M\) be diagonal matrix obtained from \(n\times n\) identity matrix I by replacing 1 by -1 in the meth row for each \(m\epsilon\) M. The matrix \(I_ M\) is called a signature matrix. The notation \(x>0\) is used for positive vectors, in each element of x is a positive real number. The main result of the paper is contained in Theorem 1: A matrix A is a P-matrix iff for each \(M\subseteq N\), \(I_ M\) A \(I_ M\) \(x>0\) has a solution \(x>0\). Such a characterization of P-matrices is known [cf. \textit{P. Moylan}, Linear Algebra Appl. 17, 53-58 (1977; Zbl 0356.15005)] but the proof of the sufficient condition given in this paper seems to be new.