Soutenance de Carlos Granero Belinchon
Multiscale Information Transfer in Turbulence
| When |
Sep 26, 2018
from 10:00 to 12:00 |
|---|---|
| Where | Salle Condorcet |
| Attendees |
Carlos Granero Belinchon |
| Add event to calendar |
vCal
iCal
|
Most of the time when studying a system, scientists face processes whose properties are a priori unknown. Characterising these processes is a major task to describe the studied system. During this thesis, which combines signal processing and physics, we were mainly motivated by the study of complex systems and turbulence, and consequently, we were interested in the study of regularity and self-similarity properties, long range dependence structures and multi-scale behavior. In order to perform this kind of study, we use information theory quantities, which are functions of the probability density function of the analysed process, and so depend on any order statistics of its PDF.
We developed different, but related, data analysis methodologies, based on information theory, to analyse a process across scales τ. These scales are usually identified with the sampling parameter of Takens embedding procedure, but also with the size of the increments of the process. The methodologies developed during this thesis, can be used to characterize stationnary and non-stationnary processes by analysing time windows of length T of the studied signal.