Post-Doctoral Position for development of self-learning digital twins

A digital twin is a representation of a physical object through data, multi-scale probabilistic simulations, and mathematical equations. It can be connected to the real world via physical sensors (e.g., temperature sensors) and can support the design and engineering of a product or enables predictions about the development of a physical object. By feeding sensory data into the digital twin the physical and the virtual world are bridged. One of the fundamental challenges in using the digital twin consists in handling of uncertainties. Uncertainties can enter mathematical models and experimental measurements in various ways. The study of the behaviour of parameterized non-linear dynamic models is often impeded by lack of knowledge of a subset of parameters (uncertainty) and/or non-uniqueness (variability) of another subset of the parameters. A core topic in the field of uncertainty quantification is the question how uncertainties in model inputs are propagated to uncertainties in model outputs. Quantification of the output of the digital twin and reacting to new data by updating the digital twin resulting into improved predictions is a challenging topic.