Block Relaxation Methods in Statistics

Abstract

This book discusses block relaxation, alternating least squares, augmentation, and majorization algorithms to minimize loss functions, with applications in statistics, multivariate analysis, and multidimensional scaling. This is a work in progress which will be expanded/updated frequently and unpredictably. It will never be finished, or, more precisely, it will be finished when I am. All suggestions for improvement are welcome. At gifi.stat.ucla.edu/bras there is a pdf version of the book, the Rmd file with all the code chunks, the tex, md, and html files, the figures, the R code, and whatever else is needed for perfect reproducibility.