Some Thoughts on Gaussian Processes for Emulation of Deterministic Computer Models

Explore the application of Gaussian Process (GP) emulators in uncertainty quantification (UQ) for complex physical systems in this insightful 54-minute conference talk by Michael Stein at the Alan Turing Institute. Delve into the theoretical and numerical aspects of GP emulation, focusing on its use in approximating computationally expensive and complex computer models. Gain valuable insights into how GP emulators can replace intricate codes with simpler, more cost-effective alternatives while maintaining essential functional relationships. Discover the relevance of this approach in addressing global challenges, particularly in climate, tsunami, and earthquake modeling, where 'large' features such as complex physical models or numerous parameters are common. Learn how this powerful technique can enhance understanding of physical systems and improve decision-making under uncertainty.