Genealogy of a multilevel splitting algorithm
Lead Research Organisation:
University of Warwick
Department Name: Statistics
Abstract
Context: Multilevel splitting is an algorithm used for rare events simulation. It requires cloning the simulation at the hitting time of levels, which induces a genealogy to the set of trajectorties that reach the final region of interest.
Aims: This project explores the coalescence of terminal trajectories. We research how ideas from statistical genetics could help with the tuning of a multilevel splitting algorithm, in particular the choice and number of levels, as well as the distribution of offspring.
Novelty: The genealogy of a branching process with location dependent branching rates and offspring distribution has not been described previously.
Aims: This project explores the coalescence of terminal trajectories. We research how ideas from statistical genetics could help with the tuning of a multilevel splitting algorithm, in particular the choice and number of levels, as well as the distribution of offspring.
Novelty: The genealogy of a branching process with location dependent branching rates and offspring distribution has not been described previously.
Organisations
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/V520226/1 | 01/10/2020 | 31/10/2025 | |||
| 2733781 | Studentship | EP/V520226/1 | 03/10/2022 | 30/09/2026 | Julia Krasnowska |