A new Optimization Algorithm for the Design of Integrated Circuits

Abstract

This paper presents a new hybrid algorithm for global optimization of integrated circuits. The algorithm exploits the efficient search mechanism of differential evolution and good global search capabilities of simulated annealing, while avoiding their weaknesses. It is easy to implement and has only a few parameters. The performance of the algorithm is verified on seven real-world cases of integrated circuit design with promising results. The proposed algorithm was implemented in SPICE OPUS simulation and optimization tool and compared with a multistart version of the constrained simplex algorithm. It outperformed the latter in terms of the final solution quality and speed.