Date: Wednesday August 27, 2014, at 3PM
Location: Cheek Hall 204
Professor Jeremy Levesley, from the University of Leicester, UK will be presenting a talk on “Multilevel sparse grid approximation using anisotropic Gaussian kernels“. This colloquium is part of the MSU Mathematics Lecture Series. Everyone is invited to attend.
Abstract: Theoretically, radial basis functions are good tools for high dimensional approximation. However, there has been little done for concrete algorithms in higher than three dimensions. In this talk, I will describe a sparse grid algorithm using a radial basis function which has been implemented in 10 dimensions. It can be shown that this is an interpolatory algorithm when the basis is of tensor product form. I will outline more recent work where we consider quasi-interpolatory schemes and demonstrate that this may be a fast alternative.