Discontinuous Galerkin time domain method for diode modeling

Summary form only given. Discontinuous Galerkin time-domain (DGTD) methods provide an accurate high-order solution of Maxwell´s eqations. In general, DG methods decompose the computational domain into a number of disjoint polyhedral elements. For each polyhedron, electromagnetic (EM) fields are evaluated as the linear combination of the set of local basis functions. Across element interfaces, the tangential continuity of fields are weakly enforced through the so-called numerical flux. Being discontinuous, DG methods support various types and shapes of elements, non-conformal meshes and local choice of basis function type and order. Further, the resultant mass matrix is a block diagonal matrix, thus it can lead to a fully explicit time-marching scheme in time-domain solution. And this leads to high efficiency in parallelization. Compared with frequency-domain method, time-domain method can characterize the broadband properties through a single simulationl thus it is suitable for modeling and capturing non-linear phenomenon. With the flexibility and capability that it provides, DGTD method is wellsuited for complex modern electronic circuit modeling, which involves multiphysics phenomenon and multi-scale structure and thus demands sophisticated and efficient numerical simulation tools, with the ever-increasing operating frequencies and the constant shrinking packaging sizes. In this paper, DGTD is proposed to model non-linear circuits networks, particularly, the incorporation of diode modeling as lumped element. The basic I-V relationship is used to describe the lumped element surface in the circuit part; the coupling between the EM and circuit parts happen at the lumped surfaces where the voltage and current on the surface are computated. This process only involves small coupling systems since DGTD method has the local properties. To further increase the efficieny, a local time-stepping method is applied for time-marching scheme.