Mathematical Diseases in Climate Models and How to Cure Them

Explore a 50-minute conference talk that delves into the challenges of climate modeling and innovative approaches to improve prediction accuracy. Learn about the use of the Julia programming language and GPUs in developing a fast, user-friendly climate model. Discover how researchers are addressing the limitations of traditional Fortran-based models and leveraging high-resolution simulations to capture small-scale physics. Gain insights into the complexities of simulating climate systems, the potential of embedding GPU-accelerated simulations within global models, and the application of machine learning techniques to enhance climate predictions. Follow the discussion on mathematical issues in climate and ocean models, parametrization techniques, neural networks, and programming language considerations for climate science.

Introduction
Mathematical diseases in climate models
Mathematical diseases in ocean
Parameterization
Neural Network
Programming languages
Kernel fusion
The perfect solution
Most specific method
Type inference
Questions answers