Back to News List
Uncertainty Quantification in KIT
Scientific Computing Center of KIT.
When and Where?
2026-02-01 to 2026-04-30, Karlsruhe, Germany
What I have done?
Scientists use models to simulate and understand the world, whether through physics-based approaches or data-driven methods.
However, as models become more complex and involve parameters that are difficult to measure,
it becomes increasingly challenging to assess the uncertainty and robustness of their outputs.
Instead of treating a model as a purely deterministic process, it can be more appropriately framed as a probabilistic one.
Thanks for the fudning of
HIDA Mobility Program
, I get the chance to work with
Prof. Dr. Sebastian Krumscheid,
who leads a Uncertainty Quantification
(UQ)
research group in KIT.
During this three-month project, I applied Markov Chain Monte Carlo (MCMC) methods to integrate historical observations and prior knowledge,
enabling the quantification of uncertainty in the stTwin framwork.
The inferred posterior parameter distributions were then used to improve model predictions under new climate forcing scenarios.
We hope to share our manuscript soon.