Return to Colloquia & Seminar listing
Best Matrix Decompositions
Special EventsSpeaker: | James Bremer, Yale University |
Location: | 1147 MSB |
Start time: | Mon, Feb 12 2007, 4:10PM |
We will discuss adaptive matrix decompositions which generalize the Best Basis algorithm of Coifman and Meyer. The idea is to select from among a library of decompositions of a matrix, the decomposition which minimizes a given cost function. The result is a matrix decomposed into malleable pieces. Depending on the choice of the cost function, these decompositions can be tailored to solve a number of different problems.
We will concentrate on a particular application: the rapid application of integral operators. We will also discuss the relationship to Fast Multipole algorithms and HSS matrix representations.
(This is joint work with R.R. Coifman.)