Learning locally dominant force balances in active particle systems

no code implementations • 27 Jul 2023 • Dominik Sturm, Suryanarayana Maddu, Ivo F. Sbalzarini

We use a combination of unsupervised clustering and sparsity-promoting inference algorithms to learn locally dominant force balances that explain macroscopic pattern formation in self-organized active particle systems.

Inverse-Dirichlet Weighting Enables Reliable Training of Physics Informed Neural Networks

no code implementations • 2 Jul 2021 • Suryanarayana Maddu, Dominik Sturm, Christian L. Müller, Ivo F. Sbalzarini

We characterize and remedy a failure mode that may arise from multi-scale dynamics with scale imbalances during training of deep neural networks, such as Physics Informed Neural Networks (PINNs).

STENCIL-NET: Data-driven solution-adaptive discretization of partial differential equations

no code implementations • 15 Jan 2021 • Suryanarayana Maddu, Dominik Sturm, Bevan L. Cheeseman, Christian L. Müller, Ivo F. Sbalzarini

Often, this requires high-resolution or adaptive discretization grids to capture relevant spatio-temporal features in the PDE solution, e. g., in applications like turbulence, combustion, and shock propagation.