Algorithm for Locating the Extremum of a Multi-Dimensional Constrained Function and Its Application to the PPPL Hybrid Study.

摘要

A description is presented of a general algorithm for locating the extremum of a multi-dimensional constrained function. The algorithm employs a series of techniques dominated by random shrinkage,steepest descent,and adaptive creeping. A discussion follows of the algorithm's application to a ''real world''problem,namely the optimization of the price of electricity, P/sub eh/,from a hybrid fusion-fission reactor. Upon the basis of comparisons with other optimization schemes of a survey nature,the algorithm is concluded to yield a good approximation to the location of a function's optimum. (ERA citation 03:038700)