On collocation matrices for interpolation and approximation (Q390630)

The authors study kernel based and other, more general collocation matrices, and their properties when parameters go to limits. In the latter case the matrices become singular in the limit. These matrices have immediate and well-known applications in the case when functions or data are interpolated from finite-dimensional spaces (they are required for the computation of the interpolation coefficients), and they also appear in the other cases of linear regression and regularization for example.