Thursday, September 21

MS7
Optimization of Engineering Systems Governed by Differential Equations - Part I of II

10:30 AM-12:30 PM
Mt. Vernon

For Part II, see MS13

High-fidelity or high-resolution mathematical models or simulations,such as the Navier-Stokes equations of aerodynamics, are designed to approximate the physical phenomena they describe to a high degree of accuracy. However, the expense of their use in iterative procedures, such as optimization, presents one of the main obstacles to applying traditional nonlinear programming techniques in engineering design. Current research aimed at addressing the computational expense of using differential equations as function evaluators in design has two major components: constructing models for computationally intensive optimization and rigorous management of the resulting variable-fidelity models in optimization. The speakers in this minisymposium will discuss both components.

10:30-10:55 Engineering Optimization with Variable-Fidelity Models

Natalia M. Alexandrov, Organizer

11:00-11:25 A Blackbox Reduced-Basis Output Bound Method for Partial Differential Equations: Application to Design and Optimization

Anthony T. Patera and Dimitrios Rovas, Massachusetts Institute of Technology, USA; Luc Machiels, Lawrence Livermore National Laboratory, USA; and Yvon Maday, Université Pierre et Marie Curie, Laboratoire d'Analyse Numerique, France

11:30-11:55 Application of a Posteriori Finite Element Output Bounds in Optimization

Luc Machiels, Lawrence Livermore National Laboratory, USA

12:00-12:25 Levels of Discretization and the Optimization of Systems Governed by Differential Equations

Robert Michael Lewis, ICASE, NASA Langley Research Center, USA