Stylized purple and blue honeycomb geometric pattern representing geometric deep learning.

Overview Resources

Altair® physicsAI™ Geometric Deep Learning

Better Design Insights Up to 1000x Faster than Solver Simulation

Altair® physicsAI™ delivers fast physics predictions by learning from your historical data without the limits of parametric studies. Accessible through Altair® HyperWorks®, our design and simulation platform, this AI-powered CAE technology trains models using any existing simulation studies including those from older design concepts, similar parts, or different programs. With its modern, geometric deep learning capability, physicsAI identifies the relationship between shape and performance for any physics. Once trained, physicsAI models can deliver predictions up to 1000x faster than traditional solver simulations, enabling teams to evaluate more concepts and make better design decisions.

Overview Video

Why PhysicsAI?

Accelerate Design Cycles

physicsAI operates directly on mesh or CAD models to produce fully animated physics outcomes at blazing speed across diverse physics applications. This streamlined approach takes a fraction of the time needed for traditional solver simulation and offers invaluable design performance insights. Whether you’re working with a crash model or HVAC design, physicsAI predictions slash runtime to seconds and what-if simulation studies from months to days.

Lightbulb that represents innovation is shining in a gray box-like room in front of a wall filled with math equations.

Innovate Faster

physicsAI delivers fast physics predictions that enable engineering teams to test more design variations than possible with traditional solver simulation alone. More design exploration in less time helps companies discover ways to improve designs early in the development cycle so that they can bring innovations to market faster than the competition.

Woman is looking at a computer monitor that is displaying simulation results generated from physicsAI.

Predict with Confidence

AI-powered technology leverages your historical data to deliver the best possible physics predictions. At the training stage, powerful geometric deep learning trains physicsAI models with your specified simulation data, regardless of the data’s origin. To ensure reliable predictions, the physics AI environment provides workflows to assess predictions and validate them against traditional solver simulations.

Key Features

Native CAE File Support

The solver-agnostic physicsAI modeling environment lets you work directly with native CAE models, including historical simulation data models.

Geometric Deep Learning

physicsAI models are trained with groundbreaking geometric deep learning that operates directly on meshes and CAD models. This technology eliminates the time-consuming hand-crafting of parameters needed with other training methods.

Confidence Score Metric

physicsAI offers a confidence score that helps identify novel shapes in your data. By scoring geometric similarity, physicsAI prevents questionable predictions and ensures reliable results.

Altair HyperWorks-Guided Workflows

Simple workflows let you select trained models, generate predictions, and assess quality on a broad range of physics such as computational fluid dynamics (CFD), crash, and manufacturing.