| | Field | Response | | |
| |:-------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | |
| | Intended Task/Domain: | Automotive Aerodynamics | | |
| | Model Type: | Decomposable Multi-scale Iterative Neural Operator (DoMINO) | | |
| | Intended Users: | Automotive engineers and researchers accelerating Computational Fluid Dynamics (CFD) predictions using AI. | | |
| | Output: | Tensor (surface pressure and wall shear stress; volumetric velocity, pressure, and turbulent viscosity fields). | | |
| | Describe how the model works: | DoMINO uses three sub-networks sharing a common geometry encoding: a Global Geometry Representation Network projects the input point cloud onto a structured latent grid via multi-scale point convolutions augmented with SDF features; a Local Geometry Representation extracts subregion features from the grid around computational stencils; and a Basis Function Aggregation Network predicts and aggregates solution fields at surface and volume points via inverse distance weighting. | | |
| | Name the adversely impacted groups this has been tested to deliver comparable outcomes regardless of: | Not Applicable | | |
| | Technical Limitations & Mitigation: | The model may perform poorly for vehicle geometries significantly different from the training data. Increased errors are observed near complex flow regions (wheels, mirrors, wake). Oscillatory behavior may occur for designs with small successive geometric changes. Edge artifacts may appear near volumetric bounding box boundaries. | | |
| | Verified to have met prescribed NVIDIA quality standards: | Yes | | |
| | Performance Metrics: | Surface and volume prediction relative L1 error; drag coefficient R² (0.96 on DrivAerML test set) | | |
| | Potential Known Risks: | This model may inaccurately predict aerodynamic fields for vehicle designs outside the training distribution. | | |
| | Licensing: | Use of this model is governed by the [NVIDIA Open Model Agreement](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-agreement/). | | |