A novel design space reduction method for efficient simulation-based optimization

Simulation-based optimization and equation-based optimization are two popular approaches for analog design optimization. In order to combine the advantages of both approaches, a novel simplex-based design space reduction method is proposed in this paper to improve the efficiency of simulation-based optimization. By using the proposed two-stage optimization approach, the design space exhibiting a global optimal point can be simplified to a small region rapidly, and the simulation-based optimization can then be applied to search the region to identify the global optimal point accurately. As revealed in the experimental results, in several analog circuits, the proposed method exhibits higher efficiency than traditional global search approaches do, without sacrificing accuracy.