Combining Quantum and AI for Accelerating CFD Simulations - Part 1
Simulating fluid dynamics is central to aircraft design, weather prediction, energy systems, and many industrial processes. These systems are described by high-dimensional PDEs, which are expensive to solve and often slow, even with powerful computing resources.
Recent advances in physics-informed AI introduced physics-informed neural networks (PINNs). They can speed up simulations or fill in gaps where classical solvers struggle. At the same time, quantum computing offers new ways to represent complex problems more efficiently as their dimensionality grows.
This workshop brings these ideas together. When we combine quantum methods with PINNs, we get Quantum PINNs—a hybrid approach that is still early, but already showing promising results for challenging simulation tasks.
We'll introduce these concepts step by step and show how they can be used to accelerate simulations, reduce computational cost, or test scenarios that are too expensive to run with traditional CFD solvers.
We'll use two open-source tools: PennyLane for quantum models and NVIDIA PhysicsNemo for classical PINNs.
We will cover:
How physics-informed models solve PDEs (partial differential equations)
How training works and why physics-based loss terms matter
How hybrid quantum-classical methods fit into the pipeline
You'll see simple implementations, attention-based extensions, and hybrid quantum-classical loops.
By the end, you'll know how AI-based simulation models work, when physics-based approaches are beneficial, and how emerging quantum techniques could fit into real-world CFD and scientific computing workflows.
About the speaker:
Aditya Seshaditya is a Data Scientist with a background in Quantum Computing, Digital Twins, and AI.
DataTalks.Club is the place to talk about data. Join our Slack community!