Return to Colloquia & Seminar listing
TBA
Student-Run Research SeminarSpeaker: | Prof. Javier Pen~a, School of Industrial Administration, Carnegie Mellon University. |
Location: | 593 Kerr |
Start time: | Mon, Nov 6 2000, 2:10PM |
The condition number of a square matrix plays a crucial role in numerical linear algebra. It is a fundamental concept in perturbation theory of linear equations. It also serves as a natural parameter in the study of numerical stability and complexity of algorithms for solving large systems of equations.
The notion of condition number can be extended to the more general context of convex programming. In this talk we show how the general condition number preserves fundamental features of traditional condition numbers. In addition, we discuss the relevance of condition numbers to properties of convex programs, especially in connection with interior-point methods.