Jahrestagung der Deutschen Mathematiker-Vereinigung 2006
Rheinische Friedrich-Wilhelms-Universität Bonn
Menü
Minisymposium 7 - Stochastic algorithms and Markov processes
Organisatoren:
Prof. Dr. Andreas Eberle
Institut für Angewandte Mathematik
Universität Bonn
Wegelerstraße 6
53115 Bonn, Germany
Prof. Dr. Peter Eichelsbacher
Ruhr-Universität Bochum
Fakultät für Mathematik
44780 Bochum, Germany

Prof. Dr. Matthias Löwe
Westfälische Wilhemsuniversität
Institut für Mathematische Statistik
Einsteinstr. 62
48149 Münster, Germany

Markov chain Monte Carlo methods, sequential Monte Carlo methods, and other related stochastic algorithms have become widely used tools in many application fields of mathematics. Despite their massive use, the theoretical understanding of convergence properties of these algorithms is often rather rudimentary. This is in particular the case in high dimensional models that typically arise in many applications. Whereas formerly, research has often been carried out more or less independently in probability theory, stochastic analysis, and statistical mechanics on the one side, and theoretical computer science, discrete mathematics, and numerical analysis on the other side, recently there is a rapidly growing activity at the borderline of the different disciplines. Besides classical probabilistic techniques (e.g. martingale methods), concepts from statistical mechanics (phase transitions, critical slowing down), and techniques from stochastic analysis (decay to equilibrium of Markov processes, spectral gap estimates) as well as infinite dimensional analysis (e.g. logarithmic Sobolev inequalities) have become crucial.


Auszug zu diesem Minisymposium aus dem Programmheft (Stand: 15. Juli 2006). Weitere nützliche Informationen rund um die Tagung können der verkürzten Ausgabe des Programmheftes entnommen werden.

Programm (Stand: 07.09.2006):

Montag HS XV, Hauptgebäude, Regina-Pacis-Weg
14:30-15:30 Olle Häggström (Göteborg)
Problem solving is often a matter of cooking up an appropriate Markov chain
15:30-16:00 Nikolaus Schweizer (Bonn)
Local spectral gaps on the mean field Ising model and Multilevel MCMC methods
16:00-16:30 Emilio de Santis (Roma)
Exact sampling for discrete time spin systems and unilateral fields
16:30-17:00 break
17:00-18:00 Pierre del Moral (Nice)
Coalescent tree based functional representations for some Feynman-Kac particle models
18:00-18:30 Andreas Eberle (Bonn)
Convergence of sequential MCMC methods I
18:30-19:00 Carlo Marinelli (Bonn)
Convergence of sequential MCMC methods II
Dienstag Großer Hörsaal, Mathematisches Institut, Wegelerstr. 10
14:30-15:30 Dan Crisan (Imperial College London)
Solving the filtering problem in a continuous time framework. Advantages and pitfalls
15:30-16:00 Wilhelm Stannat (Darmstadt)
On stability of the optimal filter
16:00-16:30 Samy Tindel (Nancy)
A model of Brownian directed polymer in a Gaussian random environment
16:30-17:00 break
17:00-18:00 Mark Jerrum (Edinburgh)
Tight bounds on mixing time of Markov chains
18:00-19:00 Marek Karpinski (Bonn)
Metric construction for path coupling, rapid mixing, and approximate counting
Mittwoch Großer Hörsaal, Mathematisches Institut, Wegelerstr. 10
14:30-15:30 Wilfrid S. Kendall (Warwick)
Perfect simulation: a survey and recent developments
15:30-16:30 Marie-Colette van Lieshout (Amsterdam)
Perfect simulation for length-interacting polygonal Markov fields in the plane
16:30-17:00 break
17:00-18:00 Rudolf Grübel (Hannover)
Markov chains in the analysis of algorithms
18:00-18:30 Ralph Neininger (Frankfurt)
The size of random fragmentation trees