A finite algorithm for a class of nonlinear multiplicative programs (Q1422889): Difference between revisions

This paper discusses the existence of solutions and the generalized convexity of the objective function to the problem \[ \min_{x\in X}\,(\textstyle{{1\over 2}}\langle x,Qx\rangle+ \langle q, x\rangle+ q_0)(\langle d,x\rangle+ d_0)^p, \] where \(X= \{x\in{\mathbb R}^n: Ax\geq b\}\), with \(A\in{\mathbb R}^{m\times n}\), \(q,d\in{\mathbb R}^n\), \(b\in{\mathbb R}^n\), \(p\), \(q_0\), \(d_0\in{\mathbb R}\), \(p\neq 0\), \(Q\in{\mathbb R}^{n\times n}\) is symmetric and positive definite and \(\langle d,x\rangle+ d_0> 0\,\forall x\in X\). The study is carried out for all \(p\in{\mathbb R}\), \(p\neq 0\). The unifying approach developed in this paper uses the so-called ``optimal level solutions'' method found in the literature and applied to solve various classes of optimization problems. In addition, a finite algorithm which solves the above-mentioned problem (having possibly unbounded feasible set) is proposed.