Description de proposition

Quasi-linear differential-algebraic equations is a convenient model structure for dynamical systems. Such models generally contain both exactly known and uncertain coefficients. The structure that the exactly known coefficients adds to the equations can be utilized when analyzing or solving the equations, but requires a more detailed model structure than general DAE. When developing theory and/or software for DAE, one needs to decide what additional structure to impose on the equations, and it is tempting to take on as general forms as possible, both to gain wide applicability and to avoid the extra bookkeeping that any additional structure would require.

In this talk, the consequences of neglecting the structure that exactly known coefficients bring to the equations will be discussed. The resulting problem will also be motivated by symbolic-numeric approaches to integration of exactly known systems, and a general approach to tackle it will be presented. In view of available results, it is also motivated to question the use of unstructured DAE as a model structure.

First presenter		Co-presenter(s)
Name :	Henrik Tidefelt *	Name:
E-mail:		E-mail:
Affiliation:	Linköping University	Name:
Department:	Department of Electrical Engineering	E-mail:
City:	Linköping	Name:
State/Province:		E-mail:
Country:	Sweden	Name:
Talk Number:	13-01	E-mail:
Session:	13- Symbolic and numeric approaches to dynamical modeling and simulation	Schedule: Room:	Thursday, 14:30 B-4404
Related website:
Title of presentation:	Unstructured matrix-valued singular perturbations -- tackle or avoid? Slides of this talk (pdf)
Abstract:
Slides of this talk (pdf) Quasi-linear differential-algebraic equations is a convenient model structure for dynamical systems. Such models generally contain both exactly known and uncertain coefficients. The structure that the exactly known coefficients adds to the equations can be utilized when analyzing or solving the equations, but requires a more detailed model structure than general DAE. When developing theory and/or software for DAE, one needs to decide what additional structure to impose on the equations, and it is tempting to take on as general forms as possible, both to gain wide applicability and to avoid the extra bookkeeping that any additional structure would require. In this talk, the consequences of neglecting the structure that exactly known coefficients bring to the equations will be discussed. The resulting problem will also be motivated by symbolic-numeric approaches to integration of exactly known systems, and a general approach to tackle it will be presented. In view of available results, it is also motivated to question the use of unstructured DAE as a model structure.