Disorder and critical phenomena through basic probability models. École d'Été de Probabilités de Saint-Flour XL -- 2010. (Q545395)

The main purpose of the book is to explore the effect of disorder in the framework of equilibrium statistical mechanics of lattice models, that is, for Gibbs measures with random interactions or random external fields. The book consists of eight chapters and an appendix. Chapter 1 is an introduction. In Chapter 2, a general class of homogeneous pinning models is introduced and solved. The emphasis is on the renewal process viewpoint, and the author is interested in when the process comes back to 0 and not so much in what it does outside of 0. The return times to 0 form a random sequence that is a renewal sequence. This chapter also contains a quick review of the modeling aspects. In Chapter 3 the disordered pinning models and some basic techniques are introduced. This chapter contains the notion of disordered free energy. In the next chapter, the Harris criterion is introduced and two heuristic arguments in favor of this criterion are provided. The notions of relevant and irrelevant disorder are introduced. A mathematical proof of disorder irrelevance in agreement with the Harris criterion prediction is given. In Chapter 5 it is shown that disorder is relevant when predicted by Harris criterion. Moreover, this chapter contains an overview of what is known or is expected to hold for the Ising model. In Chapter 6 the shift of the critical point is studied. This includes the analysis of the case of marginal disorder, i.e., neither relevant nor irrelevant, a debated issue in the physical literature. In the next chapter the coarse graining estimates are proved. Chapter 8 is devoted to path properties and it is showed that they are tightly linked to the free energy. The appendix is about discrete renewal processes and it is split into two parts: in the first part some basic tools and results of the theory are reviewed, the second part is about some issues that are more specific to pinning models.