While simple classical fluids are reasonably well understood, theories about the emerging physical behaviour of fluid particles that interact through an anisotropic potential are still quite limited. We are using machine learning methods, especially in the context of symbolic regression, to help us find an effective description (density functional theory) of these systems.
