Nonnegative definite matrices and their applications to matrix quadratic programming problems
Yonglin Chen · Linear and Multilinear Algebra · 1992
In this paper, the concept that a matrix is nonnegative definite over a subspace and the tool of generalized inverse are used to express a general form of matrix quadratic programming. Several fundamental conclusions are obtained. An application to the common penalty method for handling constrained minimization problem is given.