A heuristic global optimization algorithm and its application to CMOS circuit variability minimization

A global optimization method with heuristic searching methodology is presented. By introducing the strategy of cluster analysis and random reflection adjustment, the optimum searching procedure is put in an adaptive mode. Numerical testing and design practice of IC variability minimization show that the algorithm is moderately robust and efficient. Since no strict assumptions about the objective function are made, it is suitable for different formulations of various design optimization problems, such as l/sub 1/ norm and minimax designs.<
>