Title: Exact Linear Algebra
Speaker: B. David Saunders
Department of Computer & Information Sciences, University of Delaware
Abstract:
To solve a linear system is perhaps the best understood of
mathematical problems. However, this is largely because the concepts
of matrix and linear system are overly general. In fact there is not
one "solve linear system" problem, but many, depending on the
structure of the matrix and application. It remains a challenge to
compute solutions efficiently as hardware evolves, and the matter is
rich in interesting computer science and mathematics and of growing
importance to computer algebra. I will survey the history and the
state of the art, and offer a view of the road ahead.