Presentation of degree project: Solving Partial Differential Equations with Neural Networks
- Date: 24 February 2023, 13:15–14:00
- Location: Ångström Laboratory, Å64119
- Type: Course
- Lecturer: Håkan Karlsson Faronius
- Organiser: Matematiska institutionen
- Contact person: Benny Avelin
Welcome to Håkan Karlsson Faronius´ presentation of his master´s thesis with the title "Solving Partial Differential Equations with Neural Networks".
Abstract: In this presentation we will show how neural networks can be utilized to numerically approximate the solutions of partial differential equations. This presentation will introduce three such approaches; namely Physics-Informed Neural Networks, the Deep Ritz Method and Fourier Neural Operators. Apart from introducing the methods, inverse problems in partial differential equations will also be briefly explored as well as how importance sampling can be used when training neural networks to accelerate the training process