Ivi Tsantili - Uncertainty Quantification and Data Inference Algorithms: Towards the
development of realistic and efficient models for complex systems


September, 11 at 10:00 (MSC)


Probabilistic modeling allows the inclusion and quantification of uncertainty for some of the system’s components when, due to the system’s complexity and/or lack of access/knowledge to all of the involved scales/mechanisms of interactions, the information is insufficient to successfully model all the involved features using deterministic mathematical modeling. Inclusion of the latter to dynamical equations, modeling the laws of physics, enables a better understanding of how these uncertainties act and evolve in time. In this framework, methods for uncertainty quantification and data inference of the unknown quantities have become especially important during the last decades that the availability of data and computational power has grown exponentially. In this talk we are going to make an overview of some obtained results towards the development of realistic and efficient models that can capture data variability, overcome simplifying assumptions such as the Gaussian distribution, the Markovian property and the temporal and spatial separability of covariance functions. Finally, we are going to review some methods on data inference for stochastic differential equations and discuss directions for future research.