Prof. Hanns Ludwig Harney


Bayesian Inference - second revised edition

The present book, although theoretical, deals with experience. It questions how to draw conclusions from random events. Combining ideas from Bayes and Laplace with concepts of modern physics, we answer some aspects of this question.

The book combines features of a textbook and a monograph. Arguments are presented as explicitely as possible with the aid of of appendices containing lengthy derivations. There are numerous examples and illustrations, often taken from physics research. Problems are posed and their solutions provided.

The theory presented in the book is conservative in that the most widely-known Gaussian methods of error estimation remain untouched. At the same time, some material is unconventional. The non-informative prior is considered the basis of statistical inference and a unique definition is given and defended. Not only does the prior allow one to find the posterior distribution, it also provides the measure one needs to construct error intervals and make decisions.

The example of the binomial distribution - sketched in fig. 5.1 p. 56 - represents 300 years of statistics research. It was the first clearly formulated statistical model and the first example of statistical inference. We hope to convince the reader this subject is not yet closed.

Bayesian Inference

Please read the table of contents,
as well as the extract from chapter I of the book.

Bayesian Inference - second revised edition
Data Evaluation and Decisions
Harney, Hanns L.
2016, 243 p. 36 illus.,
Hardcover ISBN: 978-3-319-41642-7
eBook ISBN: 978-3-319-41644-1


Here are corrections and comments for the second revised edition of the book "Bayesian Inference", all collected in one pdf-file. Please download it, here (last update January, 5th 2018).

First Edition:

Go to original.