OR1ON-EIRA: Physics-Informed Neural Networks for Navier-Stokes Discovery — GPU+Pi5 Distributed Stack

Zenodo (CERN European Organization for Nuclear Research) 2026 Preprint Open access

Read full text
Authors: Gerhard Hirschmann, Elisabeth Steurer, OR1ON-EIRA
DOI: 10.5281/zenodo.18955076
Cited by: 0

Abstract

We demonstrate OR1ON-EIRA's capability to solve and verify 2D incompressible Navier-Stokes equations via Physics-Informed Neural Networks (PINNs) on a distributed stack: RTX 3060 GPU (CUDA 12.4) + Raspberry Pi 5 (ORIONEIRARASPBERRYPI). Re=100 Cavity Flow: continuity RMS=0.0002 (50× better than threshold), Reynolds number discovered with 0% error via inverse PINN. SINDy Lorenz: R²=0.9994. GPU speedup: 23× vs CPU. Honest capability assessment included.

Fields of Study & Subjects

Model Reduction and Neural Networks Neural Networks and Applications Computational Fluid Dynamics and Aerodynamics

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
11 March 2026
Language
EN
Document type
Preprint