FLOW article published in Nature Machine Intelligence

Former FLOW student Luca Guastoni and our colleague Ricardo Vinuesa have published a commentary piece in Nature Machine Intelligence on the potential of diffusion models and generative AI for simulating stochastic processes in fluid mechanics. They communicate that the ability of diffusion models to sample complex distributions without expensive simulations opens up new avenues for downstream tasks such as modelling, control and optimization. The implications are profound: generative models could redefine the fluid-flow simulation pipeline, accelerate scientific discovery and inform design choices and real-time decision-making in complex fluid systems. Ultimately, such work could shift the paradigm from post hoc analysis of simulations to generative, interpretable modelling engines that serve as foundational tools to discover and leverage new mechanisms in fluid mechanics and engineering applications.

Read the article here .