Most of you are probably aware of the sad events that marked last Spring: two founders of the department, Joe Hodges and Lucien Le Cam passed away. Rather than trying to summarize their contributions to our department and profession in a short space here, let me invite you to visit web pages we have set up in their memory http://www.stat.berkeley.edu/users/spector/hodges/ and http://www.stat.berkeley.edu/users/rice/LeCam/. In particular, I would like to call your attention to 27 unpublished papers of Lucien that we have scanned and put on the web site. The papers, which date from 1950 up to the present, show his wide range of interests; I'm sure you will find them interesting.
We are very pleased that Bin Yu is now back from a two year leave of absence at Bell Labs where she was involved in a spectrum of research at the interface of statistics and information technology. This will certainly be a key growth area for statistics. The rapid growth of information technology is changing the world, and while the research that has fueled this growth has traditionally been housed in university departments of Computer Science and Electrical Engineering, many of the key research issues that are now on the agenda for information technology engage the field of Statistics in fundamental ways. With decentralization of computing resources, the basic objects in information technology are no longer machines but data and information. The components of the infrastructure are required to exhibit a high degree of autonomy in complex, uncertain environments, and statistical decision-making ideas have become increasingly relevant in their design. Computers are increasing required to behave in an adaptive way, as for example in speech recognition.
This Fall, Bin offered a graduate topics course on information theory and statistics inspired by her research interests. It covered basic ideas in information theory, data compression, algorithmic complexity, frequentist and Bayesian inference, and the Minimum Description Length principle and its application to model selection. As well as our students and visitors, a number of graduate students from Electrical Engineering and Computer Science attended.
David Steinsaltz, our Neyman Assistant Professor, mounted a graduate course on the fascinating topic of the stochastic theory of evolution. Topics included random walks and genetic drift, random games and the evolution of communication and cooperation, branching processes, random dynamical systems and selection, statistical arguments in evolutionary thermodynamics, and evolutionary approaches to demography and life-history theory.
We are delighted to have Yuval Peres back from leave at Hebrew University. Together with Alistair Sinclair of Computer Science he is running a seminar on "Probability and Algorithms." In recent years, the importance of randomized algorithms, on the one hand, and developments in the study of random processes on graphs, on the other hand, have moved theoretical computer scientists and probabilists closer, to the point where many researchers in both disciplines are studying the same problems, albeit from different perspectives. (David Aldous and Persi Diaconis were among the first to recognize these connections). Notions from statistical physics, especially the concept of phase transition, are playing a crucial clarifying role, and seem applicable far beyond the traditional domain where mathematical physicists developed them. The goal of this seminar is to further interaction between researchers and students in computer science and probability.
We are thus very pleased to announce that Professor Sinclair has been appointed to a 0% FTE in the department.. His research interests have considerable overlap with those of our faculty and include the design and analysis of randomized algorithms, computational applications of stochastic processes, Monte Carlo methods in statistical physics, and combinatorial optimization. He was the thesis director of Ben Morris, a student in our department, who subsequently received an NSF postdoctoral fellowship based on that work. In a recent paper he uses a novel analysis of a Markov chain on near-perfect matchings of a graph, to obtain a polynomial-time algorithm for approximating the permanent of a zero-one matrix. This was a "benchmark" open problem in theoretical computer science for twenty years.
Our 0% FTEs, Leo Goodman (Sociology), Nick Jewell (Biostatistics), Mark van der Laan (Biostatistics), Ken Wachter (Demography) and now Alistair Sinclair are real sources of strength for the department. I regret to announce that Paul Holland who had a 0% FTE with us (100% in Education) has decided to return to the Educational Testing Service.