home · mobile · calendar · colloquia · 2002-2003 · 

Colloquium - Burl and Tufo

Research Reports
Michael Burl
Assistant Professor, Department of Computer Science
Associate Professor, Department of Computer Science
Professor Burl -- Intelligent Systems: Near and Far

New algorithms and increased computing power are enabling the development and deployment of intelligent systems that can sense, understand, and interact with their environment. A fundamental challenge in this endeavor is to reliably extract relevant symbolic "knowledge" from large volumes of sensor data. This talk will provide an overview of some of my prior work in this area, which draws upon techniques from machine learning, statistics, signal processing, and computer vision. Applications include mobile robotics, scientific data mining, detecting explosives using an "electronic nose", and enabling a spacecraft to autonomously locate and respond to planetary features.

Professor Tufo -- Terascale Spectral Element Simulations

The spectral element method is a weighted residual approach that employs Nth-order tensor-product polynomial bases (typ., N = 5-15) within K deformable subdomains. This approach combines the accuracy and efficiency of spectral methods with the geometric flexibility of finite elements. Because of its low numerical dispersion and dissipation, the method is ideal for the simulation of transitional and weakly turbulent flows. We begin with an overview of the method and then focus on the algorithms and implementations that have resulted in terascale performance (1.25 Tflops on 512 nodes of Lemieux at PSC). We conclude with a description of current research plans and results from a series of challenging problems in fluid mechanics.

Department of Computer Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
May 5, 2012 (14:13)