Simulating Physics with Graph Networks
Title: Simulating Physics with Graph Networks
DNr: Berzelius-2022-41
Project Type: LiU Berzelius
Principal Investigator: Rickard Armiento <rickard.armiento@liu.se>
Affiliation: Linköpings universitet
Duration: 2022-02-24 – 2022-09-01
Classification: 10299
Keywords:

Abstract

We apply for GPU resources to evaluate and investigate extension of the Graph Neural Network (GNN) framework GraphNets of Sanchez-Gonzalez et al. [https://arxiv.org/abs/2002.09405] for particle based simulations for physics-based animation. This network has been used to simulate a wide variety of challenging physical domains, involving fluids, rigid solids, and deformable materials interacting with one another. The project work and computations are conducted by Oskar Andersson as a diploma project, with Jose Luis Lima de Jesus Silva as supervisor, and Rickard Armiento as examiner.