Exploring System Dynamics in the Natural World with AI Lecture Series

These are the lectures from this week's Exploring System Dynamics in the Natural World with AI conference that happened in Oslo.

Some of the takeaways from the conference include:

* Convert problems into infinite dimensional problems if you can, wait for the mesh for last

* Spectral methods can solve a lot of problems for cheap

* It is possible to make a machine precision machine learning model for navier stokes

* There is an unexplored space in low dimensional chaotic problems that's possibly empty but possibly not

* It's possible to traceback a neural network's decisions back to where it bifurcated from a much different decision, in a different talk, it's possible to traceback from a generated image to the training images that were used in training that were relevant to the predicted image

* There is a time scaling issue in how AI and other technology moves from the theory and lab into application and education and that speed is variable. And in education this is quite slow.

A central goal of the conference was to encourage diversity and collaboration between different disciplines, all through AI/ML practice, and to identify new connections between applications in fields, from networks to earthquakes, and turbulence to chaos.

The speaker list is below, those with asterisks did not want their talks recorded.

Anders Malthe-Sørenssen Department of Physics, University of Oslo

*Caterina De Bacco MPI for Intelligent Systems, University of Tubingen

Ching-Yao Lai Geophysics and Computational Engineering, Stanford

Danny Caballero Physics and Computational Math, Michigan State University

Felix Kohler Expert Analytics

*Joachim Mathiesen Niels Bohr Institute

Karianne Bergen Data, Earth, and Computer Science, Brown University

Nikola Kovachki Nvidia, NYU

*Omar Ghattas Oden Institute, The University of Texas at Austin

Pia Zacharias Statkraft

William Gilpin Department of Physics, The University of Texas at Austin