The role of the choice of the physical model in the optimization of nanoelectronic device simulators

In order to obtain fast and reliable simulators for nanoelectronic devices, both an algorithmic optimization and an accurate choice of the physical model adopted to describe the device properties and behavior is necessary. In this article some different alternatives which exist at the modeling level are examined, with a few examples derived from the author's research activity, and their effect on the simulator performance and on the accuracy of the simulation results are discussed. I show how, while in many cases more detailed results can be obtained only at the expense of a larger computational cost and therefore a careful choice of the desired level of accuracy has to be made when writing a simulation code, in other situations the use of physically equivalent descriptions can determine a large variation in the complexity and efficiency of the simulation.