Simulating Complex Particle Dynamics with Graph Neural Networks
2025 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE credits
Student thesis
Abstract [en]
Particles are fundamental tools for simulating a wide range of physical systems, yet modeling complex particle dynamics remains a major computational challenge. This thesis explores graph neural network (GNN)-based simulators as a versatile solution to this problem, examining their effectiveness and practical limitations. Using the Graph Network-based Simulator (GNS), experiments demonstrate an ability to capture intricate behaviors across multiple domains, including water, sand, and viscoelastic materials, within a unified computational framework. Although achieving promising accuracy in short-term predictions, the model can develop instabilities during extended simulations, particularly when encountering out-of-distribution scenarios like boundary crossings. These findings highlight the potential of GNN-based methods for rapid prototyping and exploratory analysis while recognizing the need for improved robustness and scalability for broader scientific and industrial applicability.
Place, publisher, year, edition, pages
2025. , p. 38
Series
IT ; mTBV 25 010
Keywords [en]
Physics-based simulation; graph neural networks; particle methods; complex dynamics; scientific machine learning
National Category
Computational Mathematics Statistical physics and complex systems Artificial Intelligence
Identifiers
URN: urn:nbn:se:uu:diva-563878OAI: oai:DiVA.org:uu-563878DiVA, id: diva2:1984785
Educational program
Master Programme in Computational Science
Presentation
2025-06-23, 15:15 (English)
Supervisors
Examiners
2025-08-212025-07-172025-08-21Bibliographically approved