Computation stopping criteria for auxiliary problems of sequential unconstrained optimization. II: Penalty functions method (Q1813873)

[For part I see ibid. No. 1, 70-76 (1990; Zbl 0707.90086), Engl. translation in Mosc. Univ. Comput. Math. Cybern. 1990, No. 1, 67-72 (1990).] The paper deals with the nonlinear programming problem (NLP): maximize \(F(x)\), subject to \(f_ i(x)>0\) \((i=1,2,\dots,m)\), \(F\), \(f_ i\) being continuous with Lipschitz continuous derivatives on a compact subset of the \(n\)-dimensional Euclidean space. An effective method for solving the above problem of NLP is the penalty function method which comprises auxiliary unconstrained minimizations of augmented objective functions. Some stopping criteria are derived for solving auxiliary problems by using the gradient method.