Radial basis function interpolation with bound constraints.

David Bindel (CS, Cornell)

Radial basis function (RBF) interpolation is a popular approach to fitting a function with given values at scattered sample points. In this talk, we describe how to compute RBF interpolants with given function values at some sample points and satisfying upper and lower bound constraints at other points. Our approach is based on a constrained quadratic minimization problem that leads to a unique, parsimonious interpolant in which RBF centers appear only as they are needed to enforce an equality or inequality constraint.