Colloquia, Seminars and Conference News
Title : Numerical Multilinear Algebra
Date : October 30, 2008. (4:00 pm)
Location: MSB 118
Speaker : Prof. Charles F. Van Loan
Abstract:
Historically, matrix-based computational thinking has spread from specific application areas to the larger scientific community. Consider how the numerical PDE community powered the development of the first sparse matrix solvers, or how the control and signal processing engineers instigated the development of structured/parallel matrix computations. Likewise, I believe that the emerging field of tensor computations will advance through the development of numerical methods from application areas such as quantum chemistry, where the dimension of the data objects make it the perfect venue for inventing general multilinear methods that scale. Among the topics to be discussed are (a) the challenges of multilinear optimization, (b) "low rank" tensor approximation, (c) the idea of tensor networks, and (d) the need for block tensor data structures.
Bio:Professor Charles F. Van Loan is Joseph C. Ford Professor of Engineering at Cornell University.
[Back]