🚀 go-pugleaf

                             Dave Saunders
                Department of Computer and Information Sciences
                         University of Delaware


                 Title: Sparse Polynomial Matrix Determinants

Abstract: Determinants are important in a number of symbolic computational
problems.  For matrices with entries which are expressions in several
variables, the cost of computation can be much more than for the familiar
O(n^3) time, O(n^2) space numeric determinant.  Often the costs are
prohibitively much more, even when the determinant is a relatively small 
expression (zero, for instance).  

Sequential and parallel algorithms based on minor expansion, elimination,
and/or interpolation may
be used.  Each type is superior for some matrices.  I will discuss some 
analysis and experimental measurements comparing methods and I will present
a new variant of Bareiss' one-step elimination method which is effective 
for sparse matrices. 

                        THURSDAY, FEBRUARY 6, 1991
                           AE 214 - 4:00 P.M.
                          REFRESHMENTS 3:30 P.M.


This week's colloquium is hosted by Erich Kaltofen