Organization of computations in solving practical optimization problems

In numerically solving complicated optimization problems arising in practice, the search for an optimal solution frequently is a time-consuming multistage process, in the course of which the mathematical optimization model of the problem to be solved is simultaneously analyzed and adjusted. Proper organization of such a process essentially predetermines the success of the entire solution strategy, while the former mainly depends on the methods used to solve evolving optimization problems. This paper considers criteria for selecting adequate numerical methods and offers practical recommendations concerning the application of the so-called "reference-point" method, which provides for a straightforward and convenient organization of the entire computational process. © 1998 Plenum Publishing Corporation.