Researchers from Caltech and Purdue University reveal that they have solved a particular type of partial differential equation (PDE) in the Fourier domain using Artificial Intelligence algorithms (Neural Networks). Navier-Stokes is used to describe the motion of incompressible fluids, much more efficiently than traditional techniques (three orders of magnitude faster) and without requiring retraining. The neural network used also has an accuracy 30% better than other deep learning techniques previously used for similar tasks, which also required retraining for each type of fluid.
