| 摘要 |
A method optimizes a system. Performances of a plurality of design parameters of an objective function modeling the system are represented in a search space. A set of vertices are positioned in the search space, and each vertex has an associated performance. The model also includes a plurality of constraints. The constraints and the objective function are grouped according to a plurality of unique hierarchical levels. The group including the objective function has a highest level. The performance of a particular vertex is evaluated, in a low to high order of the levels, according to the groupings of the constraints and the objective function. The evaluated performance and the associated level is assigned to the particular vertex as soon as the evaluated performance is negative, unless the associated level is the highest level, in which case the assigned performance is evaluated from the objective function to partition the search space into a plurality of partitions according to the levels.
|
