IBM and Dallara use physics-based AI to accelerate aerodynamic design
IBM and Dallara reported that a Gauge-Invariant Spectral Transformer surrogate model identified the same optimal race-car diffuser design as computational fluid dynamics while producing an evaluation in about ten seconds instead of several hours. The partners plan to add more vehicle data and explore quantum or quantum-classical methods for higher-fidelity simulation.
Why it matters: Physics-based AI can cheaply screen many engineering configurations and reserve expensive simulation or experiments for the most promising candidates.