CFD and engineering design
Surface and volumetric flow prediction on complex automotive and aerospace geometries, with generalization across designs and operating conditions and learned models for optimization and inverse design.
NVIDIA Research · Learning and Perception Research
Research group
AI-Aided Engineering
Accelerating Scientific Discovery
We develop neural operators and related learning methods for physical systems. Our research spans fluid dynamics and engineering design, weather, and materials.
Why AI-Aided Engineering
High-fidelity simulation is central to modern engineering, but a single analysis can require days or weeks of computation. This limits how many designs, operating conditions, and hypotheses engineers can explore. AIE studies learning-based methods that complement numerical simulation through faster prediction, uncertainty quantification, optimization, and inverse design.
We develop learning methods for physical systems from observational and simulation data, with an emphasis on accuracy, reliability, and generalization beyond the configurations seen during training.
Methods
Our research focuses on the algorithms and model architectures required to make these models accurate, reliable, and useful. A central challenge is to move beyond treating a particular discretized mesh as the model's only representation.
We develop neural operators and related methods that can represent engineering geometry—and native CAD where appropriate—while conditioning on boundary and initial conditions, forcing, material properties, and other physical parameters. Our work focuses on accuracy, stability, uncertainty, generalization across discretizations, resolutions, geometries, operating conditions, and physical regimes.
-
Neural operators
Learning maps between functions and physical fields, with consistent behavior across discretizations and resolutions.
-
Geometry and physical conditioning
Representing complex domains, native CAD, meshes, point sets, and geometric variation while encoding boundary and initial conditions, forcing, operating conditions, and physical parameters.
-
Probabilistic modeling
Modeling uncertainty and multiple plausible physical outcomes, with evaluation based on calibration and proper scoring rules.
-
Optimization and inverse design
Using learned physical models to infer parameters, explore design spaces, and optimize engineering systems.
Research Directions
Weather
Multiscale atmospheric dynamics and probabilistic medium-range forecasting.
Materials
Mechanical and thermal response under variations in geometry, material properties, loads, and boundary conditions.
Highlighted Work
Atlas
Atlas is a scalable framework for probabilistic medium-range weather forecasting. It learns multiscale atmospheric dynamics using a compact latent representation together with a local projector for high-resolution physical fields, and supports several probabilistic estimators. Atlas was developed through a broad collaboration across NVIDIA.
NeuralOperator
NeuralOperator is an open-source PyTorch library for developing, training, and evaluating neural operators. It provides model implementations, data processing, training utilities, and examples for learning solution operators of partial differential equations and other physical systems.
Publications
-
2026
Demystifying Data-Driven Probabilistic Medium-Range Weather Forecasting
arXiv preprint
-
2025
Score-Based Diffusion Models in Function Space
Journal of Machine Learning Research
-
2024
A Library for Learning Neural Operators
arXiv preprint
-
2023
Neural Operator: Learning Maps Between Function Spaces with Applications to PDEs
Journal of Machine Learning Research
Team
AIE works closely with collaborators across NVIDIA. Publications and project pages recognize the full contributor lists for each body of work.
Interns
Dawon Ahn
Boyuan Yao
Hojjat Kaveh
Latest Updates
Group news
Introducing NVIDIA Research’s AI-Aided Engineering Group
AI-Aided Engineering is now a research group within Learning and Perception Research at NVIDIA Research, developing learning methods for physical systems.