Abstract:This paper presents some combinatorial algorithms for global optimization problems, which can be applied to solve model parameter extraction in bipolar junction transistors (BJT). The main advantage of combinatorial algorithms to a traditional algorithm like Gauss-Newton method, lies in that it needs no computation of the gradient of the objective function. Moreover, the global optimality of these approach is also better than Gauss-Newton method. As the selection of the initial point and iterative strategies proposed in this paper are based on a global search, the combinatorial algorithms are more simple, efficient, and suitable for global optimization. Examples are given to demonstrate the validity and efficiency of the proposed approach

Combinatorial Algorithms for BJT Model Parameter Extraction