ATG seminar series presents
Practical Trust-Region Methods with Optimal Complexity
by Daniel Robinson from Johns Hopkins University
Time: Wednesday, May 20 at 14:30 in 205.
I present a practical trust-region algorithm for unconstrained nonlinear optimization that also has optimal complexity. This is the first method of this type, and closes the theoretical gap that existed between trust-region algorithms and cubic regularization methods. I will focus on the new insights that have been gained as to why traditional trust-region algorithms, although efficient in practice, are unable to obtain the optimal complexity enjoyed by cubic regularization methods. To obtain the optimal complexity, our trust-region method has several new design features that emerged from our improved understanding of why conventional trust-region methods fail to achieve optimal complexity.
About the Speaker:
Daniel P. Robinson received his Ph.D. from the University of California at San Diego in 2007. He spent the following three years as a Postdoctoral Researcher in the Mathematical Institute at the University of Oxford and the Department of Industrial Engineering and Management Sciences at Northwestern University. In 2010 he joined the Department of Applied Mathematics and Statistics in the Whiting School of Engineering at Johns Hopkins University. His primary research area is optimization with specific interest in the analysis, design, and implementation of efficient algorithms for large-scale convex/nonconvex and complementarity problems. He is a member of the Society of Industrial and Applied Mathematics (SIAM), Mathematical Optimization Society (MOS), the American Mathematical Society (AMS), and INFORMS, in addition to being the INFORMS Vice-Chair for Nonlinear Optimization.