Multipoint Padé approximation for parametric model order reduction

Parameterized models appear in a wide range of practical applications. For instance during the design process for microsystems, complex models will be constructed which include geometric parameters. For the use in interconnect synthesis parameterized interconnect models are required. The usual model reduction allows a remarkable acceleration of timeor frequency-dependent evaluations. However, the reduced-order systems have the disadvantage that they allow only a small amount of alternatives in model variations. That is, each modification of the physical model such as a geometrical variation or changing boundary conditions requires a new model reduction. To overcome these difficulties it is required to generate systems of reduced order which contain additional parameters. We present recently developed techniques for parametric model order reduction based on Krylov subspace methods. The parameterized systems can be divided into several categories by the following properties: