The 2D Navier-Stokes equation describes how viscous fluids flow in two dimensions. This paper studies a version of that equation in a bounded region where the fluid is stirred by a random force, but with a twist: the randomness is not applied directly to the fluid everywhere, but is instead confined to a small subdomain and reaches the fluid indirectly through a separate heat equation component. Think of it like heating one corner of a room and watching how temperature differences eventually drive air currents throughout the entire space. The mathematical challenge is showing that this indirect, spatially limited randomness is still enough to meaningfully influence the entire fluid.
The main result is a property called exponential mixing. This means that no matter how the fluid starts moving, its statistical behavior converges rapidly and exponentially fast to a unique long-term statistical equilibrium. In physical terms, the fluid "forgets" its initial conditions quickly. Proving this is hard precisely because the noise is confined and reaches the fluid only indirectly, creating a kind of degeneracy where it is not obvious the randomness spreads enough to control the full system.
To overcome this challenge, the authors combine two sophisticated mathematical tools. The first is Malliavin calculus, a framework for doing calculus with respect to random noise that is useful for proving that probability distributions are well-behaved and spread out. The second is PDE control theory, which provides techniques for steering partial differential equations toward desired states using carefully designed inputs. The key technical contribution is a method that translates control strategies designed for the Navier-Stokes equation into ones that are compatible with the smoother, time-regular dynamics of the heat equation, allowing the two mathematical approaches to work together coherently.