Mixing, dynamics and probability
Many interesting natural phenomena are hard to predict.
When modelled as a dynamical system, this unpredictability is often the result of rapid separation of nearby trajectories.
Viewing the dynamics as acting on a probability measure, the mixing property states that two measurements (or random variables), evaluated at increasingly separated times, become independent in the time-separation limit.
Thus, the later measurement becomes increasingly difficult to predict, given the outcome of the earlier measurement.
If this approach to independence occurs exponentially quickly in time, one can profitably use linear operator tools to analyse the dynamics.
I will give an overview of these techniques and show how they can be applied to answer mathematical questions, describe observed behaviour in fluid mixing, and analyse models of the ocean and atmosphere.