Library for using the models trained in PhysicsNeMo in Engineering and CFD workflows
-
Updated
Aug 18, 2026 - Jupyter Notebook
Library for using the models trained in PhysicsNeMo in Engineering and CFD workflows
🚀 A high-performance C++ library for robust mesh input/output across numerous formats. This project serves as the ultimate "Swiss Army knife" for mesh manipulation workflows, providing blazing-fast and reliable data handling.
Official repository for the SwAIther-Precip paper.
Physical AI exploration by a CAE engineer: MeshGraphNet, FNO, PyG experiments. Companion code for the "AI Demystification" series.
Native Apple Silicon Core ML port of NVIDIA PhysicsNeMo & Earth-2 Physical AI foundation models on 16-Core ANE (38 TOPS)
NVIDIA PhysicsNeMo plugin for DimensionLab Frontier: logging, model registry, checkpoint compatibility and dataset adapters. Installable independently of the platform.
Predictive urban flash-flood engine — agentic AI for civil protection (NVIDIA Open Models Codefest 2026)
Reconstructing high-resolution fluid simulations from coarse inputs with EDM diffusion and an FNO backbone on NVIDIA PhysicsNeMo.
Frontier Python SDK: offline-first experiment tracking for physics ML, the native Rust core, and the NVIDIA PhysicsNeMo bridge.
Open-data tokamak digital twin: replay real MAST shots beside a physics model, PhysicsNeMo surrogates with honest holdout errors, NVIDIA Warp field lines, OpenUSD export.
GPU physics-informed neural operators for groundwater, benchmarked against per-well gray-box ODEs (PyTorch / PhysicsNeMo / CUDA)
Himalayan GLOF digital twin
To associate your repository with the physicsnemo topic, visit your repo's landing page and select "manage topics."