SHIFT

Instant Momentum for your Physics AI Deployment

Pre-Trained Physics AI Models

SHIFT models are trained on large-scale simulation datasets generated using Luminary's platform

3 Steps to use SHIFT

Deploy a trained model in hours rather than months with Luminary + SHIFT Models

1/ Generate data
Use the Luminary notebook to quickly generate a simulation database tailored to your design space
2/ Fine-tune SHIFT Model
Select a pre-trained SHIFT model and fine tune it with your new database
3/ Perform inference
Get fast, accurate physics predictions in your design tool of choice via model inference

Why Shift

Get to market faster with insights from physics predictions in seconds

High Quality Data

  • Don’t compromise on scale with thousands of samples per model
  • Deep simulation expertise informs data generation and validation
  • Co-developed with industry partners

Quick-start Ready

  • Get actionable insights in hours rather than months
  • Require less data to capture the design space
  • Save on model development costs

Customizable

  • Fine-tune on design-specific data for greater accuracy
  • Improve generalizability with models trained on large datasets
  • Built on modern, scalable model architecture

High Quality Data

  • Don’t compromise on scale with thousands of samples per model
  • Deep simulation expertise informs data generation and validation
  • Co-developed with industry partners

Quick-start Ready

  • Get actionable insights in hours rather than months
  • Require less data to capture the design space
  • Save on model development costs

Customizable

  • Fine-tune on design-specific data for greater accuracy
  • Improve generalizability with models trained on large datasets
  • Built on modern, scalable model architecture

Available Models and Datasets

A suite of Physics AI models and datasets

Shift - Automotive

AUTOMOTIVE

SHIFT-SUV

The first and largest foundation model for SUV aerodynamics
  • Trained on a dataset of over 10k transient DES simulations
  • Geometries generated by parametrically morphing the AeroSUV platform
  • Simulations performed using Luminary’s scale-resolving DES solver
  • NVIDIA PhysicsNeMo DoMINO model trained on the simulation dataset
Distributed under the open source CC-BY-NC license
"The Physics AI solution built with the Luminary Cloud SHIFT-SUV foundation model and Blender integration provides immediate insights into vehicle performance that designers and engineers need."
Fong Loon Pan - Honda
Fong Loon Pan
Principal Aero Design Lead, Honda
Honda - Logo
Shift-wing-illustration-imgupscaler

AEROSPACE

SHIFT-Wing

The largest open-source dataset and model for transonic wing design
  • Trained on a dataset of over 4k RANS simulations and growing
  • Geometries generated by parametrically modifying the NASA CRM
  • Simulations were performed with Luminary Mesh Adaptation to achieve high accuracy and resolve shock waves
  • NVIDIA PhysicsNeMo DoMINO model trained on the simulation dataset
Distributed under the open source CC-BY-NC license
"This partnership with Luminary Cloud will complement Otto’s ability to unlock scale in generation of Otto’s proprietary physics-based simulation data, making it possible to explore, optimize, and validate aerodynamic concepts faster than ever before."
ObiNdu-Otto
Obi Ndu
Chief Information & Digital Officer, Otto Aviation
Otto-Aviation-White - Cropped
cca

DEFENSE

SHIFT-CCA

An AI model to accelerate design and analysis of Collaborative Combat Aircraft (CCA) to ensure mission readiness
  • Trained on a dataset over 3k steady RANS simulations
  • nTop’s implicit modeling was used to generate variations of Group 5 CCA drone configuration
  • Simulations were performed on meshes generated with Luminary Mesh Adaptation to predict drone aerodynamics with high accuracy across geometries
  • NVIDIA PhysicsNeMo DoMINO model trained on the simulation dataset
SHIFT-CCA dataset and associated AI models are proprietary.
sub

MARINE

SHIFT-Submarine

An advanced physics AI model to rapidly evaluate the design space and meet critical mission requirements
  • Trained on a dataset of over 4k RANS simulations and growing
  • Geometries were generated by parametrically varying a proprietary submarine model
  • Luminary Mesh Adaptation was used simulate submarine hydrodynamics with high accuracy
  • NVIDIA PhysicsNeMo DoMINO model trained on the simulation dataset
SHIFT-Submarine dataset and associated AI models are proprietary.
pump_overlay

INDUSTRIAL

SHIFT-Pump

The comprehensive open-source dataset and model for centrifugal pump design and analysis
  • Trained on a dataset of over 5k steady-state simulations
  • Datasets are generated from varying geometric parameters and the operating conditions of a single stage centrifugal pump
  • Simulations were performed with a fixed high-fidelity mesh and Luminary’s solver to predict impeller physics
  • NVIDIA PhysicsNeMo DoMINO model trained on the simulation dataset
Distributed under the open source CC-BY-NC license
More Coming Soon

Custom Models

Develop a model tailored to your use case

Shift - Automotive

Automotive

Shift - Aerospace & Defense

Aerospace & Defense

Turbomachinery

Turbomachinery

Shift - Automotive

Automotive

Shift - Aerospace & Defense

Aerospace & Defense

Turbomachinery

Turbomachinery

Download
Access the datasets
Learn More
Read about our plan for SHIFT