Thursday, September 16, 2021 - 14:00 in ZOOM - Video Conference
Lessons from population genetics for computational statistics: genealogies of interacting particle systems used in Sequential Monte Carlo
A talk in the Keine Reihe series by
Paul Jenkins from University of Warwick, United Kingdom
| Abstract: |
Interacting particle systems are a broad class of stochastic models for phenomena arising across physics, engineering, biology, and finance. A prominent class of such models can be expressed as a 'sequential Monte Carlo' algorithm in which the aim is to construct an empirical approximation to a sequence of measures. The approximation is constructed by evolving a discrete-time, weighted population of particles, alternating between a Markov update and a resampling step. This setup is familiar in population genetics, being mathematically equivalent to a Cannings model in a moving environment. Conditions under which the genealogy of a sequence of Cannings models in a fixed environment converges to a given scaling limit are well-studied. In this talk I discuss how such results can be extended to cover a much broader class of interacting particle systems commonly used in computational statistics. More precisely, under given conditions we can show that a certain time rescaling ensures that the genealogy converges (as the number of particles grows) to Kingman's coalescent.
This is joint work with Suzie Brown, Adam Johansen, Jere Koskela, and Dario Spanò.
Registration via email to popgen.conference@techfak.de
|
Back
