Opening Exploring System Dynamics in the Natural World with AI
Exploring System Dynamics in the Natural World with AI
- Length
- 28 min
- Level
- Research
- Recorded
- UT Austin2024
Chapter 1 of 5: Before the talk
Exploring system dynamics in the natural world with AI begins by examining the role of machine learning in physical modeling and data compression. The lecture traces how neural networks function as universal approximators, contrasting naive curve-fitting with physics-informed neural networks that embed governing equations directly into the loss function. It then connects chaos theory, Lyapunov times, and Kolmogorov complexity to demonstrate how autoencoders perform data compression in latent spaces. Finally, concrete examples from fluid dynamics, chemistry, and meteorology illustrate how these methods extend predictive horizons and accelerate simulations.

Your comment is kept while you sign up.Sign in
Examines how physics-informed neural networks and autoencoders model chaos and compress complex systems.
Exploring system dynamics in the natural world with AI begins by examining the role of machine learning in physical modeling and data compression. The lecture traces how neural networks function as universal approximators, contrasting naive curve-fitting with physics-informed neural networks that embed governing equations directly into the loss function. It then connects chaos theory, Lyapunov times, and Kolmogorov complexity to demonstrate how autoencoders perform data compression in latent spaces. Finally, concrete examples from fluid dynamics, chemistry, and meteorology illustrate how these methods extend predictive horizons and accelerate simulations.
Numerical Recipes in C
The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
Multilayer Feedforward Networks are Universal Approximators
Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
Chaos: Making a New Science
Predictability: A way to characterize complexity
Model-Free Prediction of Large Spatiotemporal Chaotic Systems from Data: A Reservoir Computing Approach
Scalable molecular simulation of electrolyte solutions with quantum chemical accuracy
GraphCast paper
A probabilistic graphical model foundation for enabling predictive digital twins at scale
This lecture
Talk 8 of 8 of Exploring System Dynamics in the Natural World with AI.
Choose a node to see how it connects to this lecture. What comes before is on the leftat the top. The ring is what it teaches. People and places are above. Other lectures and books are below.
Press a chosen node again to open it.