A few weeks ago, I came across a book Dix petits démons chinois in the series Les nouvelles enquêtes du Juge Ti (The new cases of Judge Dee) by Frédéric Lenormand. As I highly enjoyed the original Judge Dee series by Robert van Gulik, I bought it. The style of this new book is rather far from the original and the detective plot is rather poor, so I do not think I will buy further books in the series. However, this started me on re-reading van Gulik’s books, which remain as enjoyable as I found them twenty-five years ago… Incidentally, a movie called Detective Dee just came out (in France) a few days ago. It again builds on the same historical Tang dynasty judge, with apparently little connection with the original books.
Archive for April, 2011
This afternoon, when working on ABC with Jean-Marie Cornuet, I came up with a domino formula for Bayes factors. It goes like this
with hopefully clear if implicit notations. I wonder if this has been exploited previously for computational purposes as each expectation is taken under the “previous” posterior: it could have an appeal from a sequential Monte Carlo perspective.
Just to remind ‘Og‘s readers that the 2011 International Workshop on Objective Bayes Methodology will take place on June 11-15th 2011 at East China Normal University (Putuo Campus), Shanghai (上海), China. The deadline for early registration has been extended till May 10. This should be a wonderful opportunity to exchange about the latest in objective Bayes methodology, to meet Chinese Bayesians and to discover that part of China. (Actually, there is a possibility of an excursion to Guilin, Xi’an and Beijing after the meeting!) Here is a picture of my visa, received today.
I have been asked to discuss the on-coming Statistical Science paper by Don Fraser, “Is Bayes posterior quick and dirty confidence?“. The title was intriguing if clearly provocative and so did I read through the whole paper… (The following is a draft of my discussion.)
The central point in Don’s paper seems to be a demonstration that Bayes confidence sets are not valid because they do not provide the proper frequentist coverage. While I appreciate the effort made therein of evaluating Bayesian bounds in a frequentist light, and while Don’s paper does shed new insight on the evaluation of Bayesian bounds in a frequentist light, the main point of the paper seems to be a radical reexamination of the relevance of the whole Bayesian approach to confidence regions. The outcome is rather surprising in that the disagreement between classical and frequentist perspectives is usually quite limited [in contrast with tests] in that the coverage statements agree to orders between and , following older results by Welch and Peers (1963). Continue reading
The book I co-edited with Kerrie Mengersen and Mike Titterington, Mixture: Estimation and Applications, has just been published by Wiley! It is a pleasure to flip through the chapters contributed by the participants to the ICMS workshop of about a year ago in Edinburgh. While there may (must) be residual typos, I did not spot any obvious mishap in the production of figures and Buachaille Etive Beag proudly stands on the cover (despite contrary advice from some ‘Og readers). It is also a pleasure to have this book published in the same series as references like Titterington, Smith and Makov’s Statistical Analysis of Finite Mixture Distributions, and McLachlan and Peel’s Finite Mixture Model. (The “product description” on amazon does not start very well, though: “This book uses the EM (expectation maximization) algorithm to simultaneously estimate the missing data and unknown parameter(s) associated with a data set. The parameters describe the component distributions of the mixture; the distributions may be continuous or discrete.” It fortunately improves by reproducing the back cover.)
Here is an email that could appeal to some readers:
Job in Cambridge MRC-BSU – Bayesian statistician
Career development fellow
MRC Biostatistics Unit, Cambridge
We are offering an exciting opportunity to work on Bayesian models for infectious disease dynamics. A statistician is required to contribute to a programme of research to develop inferential approaches to estimation and prediction of epidemic evolution when relevant information, possibly from diverse sources, arrives at regular intervals, and, likely, in real time. Examples include the monitoring and prediction of long standing epidemics such as HIV, as well as new emerging epidemics (e.g. Swine Flu). Challenges include the need to synthesise heterogeneous and biased information to produce timely updates of epidemic evolution. You will have a PhD in statistics, or a relevant discipline; computing experience, both of statistical packages (e.g. R) and programming languages (e.g. C, C++); and experience of Bayesian statistics. Experience of Bayesian computation (e.g. MCMC) and of Bayesian inference for disease transmission would be advantageous. You must have good communication skills and be able to contribute substantially to writing scientific papers. Starting salary will be in the range of £26,022 – £28,764 per annum.
Applications are handled by the RCUK Shared Services Centre; to apply please visit our job board to view the full job description and person requirements and complete an online application quoting reference IRC18709. If you are unable to apply online please contact us on 01793 867003.
Closing date: 11th May 2011 Interview date: 25th May 2011
Following a successful meeting in 2005, the DST Centre for Interdisciplinary Mathematical Sciences, at Banaras Hindu University (BHU) in Varanasi, is organizing an ISBA Regional Meeting on January 06-10, 2013. This is in almost two years from now, but it is worth advertising that early. (Besides, given the heavy traffic to and from India at this time of the year, plane tickets should be booked well in advance if not that in advance!) I have been kindly invited to this conference and I definitely plan to attend. Not only the previous program was quite exciting, but this is also the ‘city of temples’ and of ghats along the Ganges, and a place of pilgrimage for Hindus, Jains and Buddhists. Plus, this is certainly the place for a Bayesian conference that is the closest to the Himalayas!