Dr. Kalyan Perumalla
Oak Ridge National Lab
Friday, November 9, 2018
2:30-3:30 JDT 410
Agencies such as the Department of Energy are dramatically advancing the hardware capabilities for high-end computing, with Exascale computing around the corner and more to come beyond. For several applications, such leadership-class supercomputing offers unprecedented solution opportunities while also raising new challenges in algorithmic approaches. In this seminar, we focus on parallel mixed integer programming that is a versatile, abstract formulation for a gamut of combinatorial problems in operations research. To exploit certain unique architectural traits, such as accelerated nodes with multiple Graphical Processing Units (GPUs) per node, of the Summit supercomputer at ORNL, we visit the parallel branch-and-cut approach to mixed integer programming. We identify algorithmic considerations that are influenced by the parallel computing hardware idiosyncrasies and outline some of our active research directions.
Kalyan Perumalla is a Distinguished Research and Development Staff member and Manager at the Oak Ridge National Laboratory. Dr. Perumalla founded and currently leads the Discrete Computing Systems Group in the Computer Science and Mathematics Division at the Oak Ridge National Laboratory. He also serves as an Adjunct Professor in the School of Computational Sciences and Engineering at the Georgia Institute of Technology and as an ORNL Joint Professor in the Department of Industrial and Systems Engineering at the University of Tennessee, Knoxville.