Student Analysis Seminar
Relaxation enhancement for advection-diffusion equations
Speaker: Robert Trosten, NYU Courant
Location: Warren Weaver Hall 512
Date: Monday, November 2, 2026, 12:30 p.m.
Synopsis:
Consider a cup of coffee into which you've poured some milk. If we just leave this be, the concentration of milk particles will evolve according to the heat equation \(\partial_t \theta -\kappa\Delta \theta = 0\), where \(\kappa > 0\) is the molecular diffusivity. For the heat equation on a compact Riemannian manifold such as the torus, classical energy estimates tell us that the solution \(\theta_t\) relaxes to its mean as \(t\to\infty\) at a rate proportional to \(1/\kappa\), and this is as fast as you'll converge. But when we want to add milk into coffee, we don't just leave a patch of milk sitting there to diffuse — if we stir it in, the milk becomes uniformly distributed (to our eye) much faster!
Mathematically, if we introduce some stirring, we then have an advection-diffusion equation \(\partial_t \theta + u\cdot\nabla\theta -\kappa\Delta \theta = 0\) on a compact Riemannian manifold, where \(u\) is incompressible and Lipschitz, and we are interested in understanding when the rate of relaxation of solutions to their mean is meaningfully faster than the naive \(1/\kappa\) afforded by the heat equation. Such flows \(u\) are called relaxation-enhancing flows.
In this talk, we will discuss the seminal 2008 work of Constantin, Kiselev, Ryzhik, and Zlatoš, which provided a spectral characterization of relaxation-enhancing flows, proving in particular that weakly mixing flows are relaxation-enhancing.