Obtuse cones and Gram matrices with non-negative inverse (Q5947466)

Optimization methods in convex minimization problems lead to a study of the properties of a full column rank real matrix \(A\) of type \(n \times m\) whose Gram matrix (i. e. the matrix \(A^TA\)) has non-negative inverse. These properties can be used to the constructions of an obtuse cone. The constructions consist in an addition of one vector to the given system of vectors generating an obtuse cone. As a generalization of these results sufficient and necessary conditions for a block matrix \(A=[A_1, A_2]\) having a Gram matrix with non-negative inverse are presented. Finally, the results are given in terms of \(QR\)-factorization of a block matrix.