PURPOSE: To optimize a nonlinear function-linear constraint(NLFLC) system by resolving a resource assignment problem by an iteration method and executing conversion before affine scaling in each round.
CONSTITUTION: A QP (quadratic function) problem is a special case of general NLFLC problems; and though the cost function is relatively slowly changed so that it can be predicted with respect to the QP problem, the characteristic of the cost face is quickly changed at each point with respect to general NLFLC problems. For the purpose of realizing cost assignment improved under the existence of the cost function changing in this manner, linear retrieval operation traces a local optimum point in each step and is executed to ensure that the step where a minimum cost is skipped in a selected direction is not performed. Thus, the NLFLC system is optimized.
SARATO CHIYANDORASHIKAA PUTENP
RAKUSHIYUMAN PII SHINAA
| JPS59132063A | 1984-07-30 |