Simulating Physics with Graph Networks
||Simulating Physics with Graph Networks|
||Rickard Armiento <firstname.lastname@example.org>|
||2022-10-18 – 2023-05-01|
This is a minor continuation of Berzelius-2022-41 that provided resources for the diploma project "Physics-based Animation Using Graph Neural Networks" by Oskar Andersson (which was successfully defended on Aug 29 and the thesis will appear in Diva with an appropriate acknowledgement to NSC/Berzelius.) We see an opportunity that this work may be publishable as a proceedings or journal article with only a rather small extension of the resources. Hence, we ask for these resources to complete the study of exploring the Graph Neural Network (GNN) models that were implemented in Berzelius-2022-41 to simulate new types of physics; in particular the snow-like physics data set produced as part of the diploma work. Most of the computations will be conducted by Oskar Andersson, with Jose Luis Lima de Jesus Silva and Rickard Armiento supervising the work.