Linear and Nonlinear Model Order Reduction for Numerical Simulation of Electric Circuits
104 pages, year of publication: 2010
price: 36.50 €
Increasing complexity combined with decreasing geometrical sizes in electric
circuit design lead to high dimensional dynamical models to be considered by
Model order reduction (MOR) has become a popular strategy to decrease the
problem's size while preserving its crucial properties.
MOR shall achieve accurate statements on a circuit's behavior within an
affordable amount of computational time.
Just recently, MOR techniques are designed to consider the differential
algebraic nature of the underlying models.
We present an approach based on an ε-embedding, i.e., a strategy
applied in the construction of numerical integration schemes for differential
algebraic equations (DAEs).
The system of DAEs is transformed into an artificial system of ordinary
differential equations (ODEs), since MOR schemes for ODEs can be applied now.
We construct, analyze and test different strategies with respect to the usage
of the parameter ε that transforms the DAEs into ODEs.
Moreover, accurate mathematical models for MOS-devices introduce highly
As the packing density of devices is growing in circuit design, huge nonlinear
systems appear in practice.
It follows an increasing demand for reduced order modeling of nonlinear
In the thesis, we also review the status of existing techniques for nonlinear
MOR by investigating the performance of the schemes applied in circuit