# Seminar Write-up on Differentiable Physics

Project work for the Master-Seminar - Deep Learning in Physics (IN2107,
IN0014)
course at TUM in the 2022 Summer semester.

The seminar paper gives an overview of Learning to Control PDEs with Differentiable
Physics.

Compiled versions:

## Presentation

The slides of the corresponding *(originally 25 minutes)* presentation gives
a broader overview of the topic:

- Differentiable Physics
- Motivation
- Forward Physics
- Differentiable
*(“backwards”)* Physics
- Examples

- Partial Differential Equations (PDEs)
- Method and Results of the Paper
- Other Examples and Outlook
- Enhancing Simulations with Neural Networks (NNs)
- Inductive Physical Bias

As the presentation contains videos, it is also available as an online
PowerPoint. (see above)