Two differential equation systems for equality-constrained optimization (Q2383837)

The authors consider equality-constrained optimization problems. First, they modify the method of \textit{Yu.~G.~Evtushenko} [Sov. Math., Dokl. 15, 420--423 (1974; Zbl 0314.90083)]. Secondly, they use in this construction a continuous version of the Newton method. In both cases they adopt the Armijo line search rule to determine the steplengths. Appropriate convergence theorems are demonstrated; the convergence rate of the second method is locally quadratic. Some numerical methods for the constructed systems are compared by standard test examples. From the authors' abstract: ``The numerical results given here show that the Runge-Kutta method has better stability and higher precision and the numerical method based on the differential equation system with the second information is faster than the other one.''