Nonparametric goodness-of-fit testing under Gaussian models (Q1852968)

While most of the books concerning nonparametric statistics focuse on estimation theory, the present book is devoted to a modern theory of nonparametric goodness-of-fit testing. Parallel to nonparametric estimation, an asymptotic version of the minimax approach and a method of constructing asymptotically least favourable priors for a wide enough class of nonparametric hypotheses testing are the crucial elements. These methods are applied to construct asymptotically minimax tests and to obtain sharp asymptotics of minimax error probabilities for alternatives that are determined by the power and Besov norms in the sequence spaces. Jointly with wavelet transforms, they give the structure of minimax consistent tests and the rate asymptotics for alternatives determined by Sobolev and Besov norms in the functional spaces. The focus is on Gaussian models. Mainly simple null hypotheses are considered, however the main ideas and results are extended to testing of composite null hypotheses as well. The book is divided into 9 parts, 8 chapters and an appendix. Chapter 1 reviews basic notions and basic results of classical hypothesis testing. Chapter 2 contains the statements of hypothesis testing problems and known fundamental results needed in the book. The topic of Chapter 3 is minimax distiguishability. In Chapter 4, sharp asymptotics for simple asymptotics are studied while general Gaussian asymptotics are treated in Chapter 5. In Chapter 6, extreme problems in a two-sequence space and asymptotically sharp solutions for extreme problems for the power and Besov norms are obtained. Chapter 7 is devoted to adaptive hypothesis testing for power and Besov norms. Chapter 8 concerns high-dimensional signal detection. The Appendix contains some technical elements of the proofs. Chapter headings: 1. Introduction; 2. An overview; 3. Minimax distinguishability; 4. Sharp asymptotics. I; 5. Sharp asymptotics. II; 6. Gaussian asymptotics for power and Besov norms; 7. Adaptation for power and Besov norms; 8. High-dimensional signal detection. A. Appendix. Rather advanced knowledge of probability theory and mathematical statistics is assumed. The level of the book meets a quite high standard. The book will certainly be of interest to mathematical statisticians interested in the theory of nonparametric statistical inference, and also to specialists dealing with applied nonparametric statistical problems in signal detection and transmission, technical and medical diagnostics, and other fields.