Parallel Solution of Nonlinear Programming Problems on Emerging Architectures

Abstract:

Large-scale nonlinear programming (NLP) has proven to be a fundamental technology for steady state and dynamic optimization of chemical processes. However, the size and complexity of nonlinear optimization problems of interest in academia and industry continues to grow, often outstripping the capability of a single CPU workstation. Furthermore, computer chip manufacturers are no longer focusing on increasing clock speeds and instruction throughput, but rather on multi-core architectures and hyper-threading. This means that the “free” performance improvements that we have enjoyed as a result of advances in computing hardware will no longer be possible unless we develop algorithms that are capable of utilizing modern concurrent architectures efficiently. All computing architectures are not created equal, and we will discuss the key differences in parallel architectures available for scientific computing, including distributed clusters, general multicore systems, and graphics processing units (GPU). In this presentation we will discuss decomposition strategies that allow efficient solution of nonlinear programming problems on these parallel computing architectures.

Biography:Carl Laird is an Assistant Professor in the Artie McFerrin Chemical Engineering Department at Texas A&M University. He received his undergraduate degree at the University of Alberta in 2001, and his Ph.D. from Carnegie Mellon University in 2006 in Chemical Engineering. Dr. Laird's research interests include large-scale nonlinear programming and parallel algorithms for efficient solution of structured large-scale problems. Application areas include chemical process systems, homeland security, and large-scale infectious disease spread. In 2005, Dr. Laird was the recipient of the IBM Research Bravo Award for his work with IPOPT. He has been the recipient of several teaching awards and an NSF Faculty Early Development (CAREER) award.