Title :
A statistical perspective on nonlinear model reduction
Author_Institution :
Cadence Berkeley Labs., San Jose, CA, USA
Abstract :
We consider the advantages of statistically motivated reasoning in analyzing the nonlinear model reduction problem. By adopting an information-theoretic analysis, we argue that the general analog macromodeling is tractable on average by nonlinear reduction methods. We provide examples to illustrate the importance of utilizing prior information, and provide a general outline of algorithms based on reproducing kernel Hilbert space machinery.
Keywords :
Hilbert spaces; circuit simulation; information theory; nonlinear network synthesis; reduced order systems; statistical analysis; analog macromodeling; information-theoretic analysis; kernel Hilbert space machinery; nonlinear model reduction problem; statistically motivated reasoning; Circuit simulation; Circuit synthesis; Circuit testing; Computational modeling; Driver circuits; Hilbert space; Information analysis; Laboratories; Machinery; Reduced order systems;
Conference_Titel :
Behavioral Modeling and Simulation, 2003. BMAS 2003. Proceedings of the 2003 International Workshop on
Print_ISBN :
0-7803-8135-1
DOI :
10.1109/BMAS.2003.1249855
