Project III (MATH3382) 2020-21


Supercool transitions

Large deviation theory and metastability

C. da Costa

Description

Did you know you can prepare supercool fluids in your kitchen? The experiment is crystal clear and is sometimes mentioned in informal conversations. Take a bottle of pure water (other liquids might not work equaly well) and cool it down to roughly \(-5^o\)C (minus five degrees Celsius), but don't let it freeze (yes, it is possible!). Remove the bottle gently from the cold environment carefully and hit it against a rigid surface. If you are lucky, you will see, in a matter of seconds, supercool liquid water transition beautifully into ice. What just happened?

You have just seen a Metastable transition!

In a couple of words, for pure undisturbed water near \(0^o\)C, the process of freezing depends on the formation of a first crystal of water (nucleation) to be triggered. This nucleation event depends on a geometrical alignment atoms and bonding thereof that is sufficiently strong to propagate to the system, see the figure bellow for a illustration of what happened.

Potential barrier

Behind such phenomena, there is a deep and fascinating mathematical theory, that relates the study of rare events and interacting particle systems . The goal of this project is to explain this type of phenomena.

Along the way we will see that there are different approaches to follow:

  1. The pathwise approach (Olivieri-Vares),
  2. The potential-theoretic approach (Bovier-den Hollander),
  3. The martingale approach (Landim).

You will have ample freedom to investigate the topic of your preference and to explore numerical aspects of phase transition.

Prerequisites

Probability II.

Resources

  • Sick Sience Instant Freeze Water - Sick Science! #226
  • Wikipedia: Large deviations
  • Wikipedia: Interacting Particle Systems
  • Olivieri, Enzo; Vares, Maria Eulália - Large Deviations and metastability
  • Bovier, Anton; den Hollander, Frank- Meta stability a potential theoretic approach
  • Landim, Claudio- Metastable Markov chains
  • email: Conrado da Costa