| 摘要 |
A method, system and article of manufacture are disclosed for developing corner models for various classes of nonlinear systems. The method comprises the steps of determining whether an explicit relationship between one or more performance targets vs. statistical model parameters is known; and deciding, when an explicit relationship between one or more performance targets vs. statistical model parameters is known, whether the relationship is linear or nonlinear. The relationship is constructed in the fractional form when an explicit relationship between one or more performance targets vs. statistical model parameters is not known. In one embodiment, the invention provides an optimal corner model solution for a single performance target, which varies with statistical parameters nonlinearly. In another embodiment, the invention provides an optimal and common corner model solution for multiple performance targets which vary with statistical model parameters nonlinearly. A step of decreasing the order of a target function on statistical model parameters may be used in the process of generating corner models. |
