LARGE-SCALE SYSTEM OPTIMIZATION: A REVIEW

Possibilities are outlined that are believed to be of assistance in developing solution techniques for solving large-scale systems, the development of which will govern the cleverness used in instructing automated devices, the cleverness, in turn, determining the quality of the decisions made by the devices. The possibilities for developing the solution techniques include 1 taking advantage of the low density of nonzero coefficients in an original matrix, 2 finding a compact inverse representation of the basis using a simplex method, and 3 making use of the generalized linear program or decomposition principle approach. The last-mentioned technique is used for illustration on a linear-control problem where it led to the maximum principle and could be used to converge constructively to an optimal solution.