Show HN: Neural Particle Automata
Researchers have introduced Neural Particle Automata (NPA), a new framework that generalizes Neural Cellular Automata to dynamic particle systems. By using differentiable Smoothed Particle Hydrodynamics, the model enables scalable, learnable rules for complex particle-based tasks.
Why it matters
This advancement offers a more efficient and flexible way to simulate complex physical systems and morphogenesis in AI research.
Neural Particle Automata: Learning Self-Organizing Particle Dynamics Ehsan Pajouheshgar * 1 , Hyunsoo Kim * 2 , Sabine Süsstrunk 1 , Wenzel Jakob 1 , Jinah Park 2 1 EPFL, Switzerland 2 KAIST, South Korea * Denotes equal contribution SIGGRAPH 2026 arXiv GitHub Texture Demo --> Texture Demo Growing Demo --> Growing Demo arXiv --> --> --> --> --> Font Awesome fontawesome.com - >--> --> Colab --> --> --> --> --> --> --> --> Supplementary --> --> --> Equivariant Growing Demo (Bonus) This demo is not yet supported on iPhone devices . Please use desktop/laptop or Android devices.
The article is a technical summary of a research paper with no political or social bias.
Get smarter about the news
Sign up free for a feed built around what you actually care about, Dive Deeper research on any story, and the full text of every article.
Create free accountAlready have an account? Sign in