Approximation (Q5915852)

This book is an introductory work on a central subject within numerical analysis, namely approximation theory. Approximation theory studies the approximation of functions and data by simpler objects, such as other, simpler functions (examples: polynomials in one dimension, splines, i.e., piecewise polynomials etc) or expansions (e.g. Fourier series). Since the main question is about proximity, and therefore we must know how to measure it, the important questions relate to norms that are used to study the quality of approximations, convergence of approximations if concepts like decreasing spacings of data or increasing dimensions (e.g. orders of polynomials) of the spaces from which we approximate are used, and what are the best (possible) approximations should they exist. Armin Iske's book addresses all the questions in an excellent introductory and well-presented fashion, and he includes advanced themes like kernel approximations and usage of approximation theory in computer tomography. The themes are carefully introduced and also carefully presented, and they include all the fundamental theorems a beginner in the subject should know. Thus the chapters encompass best approximations, least-squares approximation, asymptotic concepts, Chebyshev approximations, and the already mentioned kernel methods, expansions and applications. This is a highly recommendable piece of work.