Brownian geometry (Q2274123): Difference between revisions

This paper is based on the 21st Takagi Lectures that the author delivered at Research Institute for Mathematical Sciences, Kyoto University on June 23, 2018. The goal of the paper is to survey a number of recent developments concerning the continuous models of planar random geometry and their connections with discrete models. The discrete models of random geometry are planar maps which are finite connected graphs embedded in the two-dimensional sphere and viewed up to orientation-preserving homeomorphisms. Starting from a random planar map \(M_n\) uniformly distributed over the class of all triangulations with \(n\) faces, one shows that the vertex set \(V(M_n)\) equipped with the graph distance rescaled by the factor \(n^{-1/4}\) converges in distribution in the Gromov-Hausdorff sense to a limiting random compact metric space which the author calls Brownian sphere, see [the author, Ann. Probab. 41, No. 4, 2880--2960 (2013; Zbl 1282.60014)]. The paper is organized as follows. Section 2 titled ``Discrete and continuous models of random geometry'' considers the Gromov-Hausdorff distance. If \(K_1\) and \(K_2\) are two compact subsets of a metric space \((E, d)\), the Hausdorff distance between \(K_1\) and \(K_2\) is defined by \(d_{\mathrm{Haus}}^E(K_1, K_2) = \inf\{\varepsilon >0: K_1\subset U_{\varepsilon}(K_2) \,\mathrm{and}\, K_2\subset U_{\varepsilon}(K_1)\}\), where \(U_{\varepsilon}(K_1)= \{x\in E: d(x, K_1)\le \varepsilon\}\) is the \(\varepsilon\)-enlargement of \(K_1\). Definition. Let \((E_1, d_1)\) and \((E_2, d_2)\) be two compact metric spaces. The Gromov-Hausdorff distance between \(E_1\) and \(E_2\) is \(d_{GH}(E_1, E_2) = \inf\{d_{\mathrm{Haus}}^E(\Psi(E_1), \Psi(E_2))\}\), where the infimum is over all isometric embeddings \(\Psi_1: E_1\to E\) and \(\Psi_2: E_2\to E\) of \(E_1\) and \(E_2\) into the same metric space \((E, d)\). Let \(\mathcal K\) stand for the set of all compact metric spaces, where as usual two compact metric spaces are identified if they are isometric. Then Gromov-Hausdorff distance \(d_{GH}\) is a metric on \(\mathcal K\) and \(({\mathcal K}, d_{GH})\) is a Polish space. A sequence \((E_n)\) of compact metric spaces converges to a limiting space \(E_{\infty}\) in \({\mathcal K}\) if and only if all spaces \(E_n\) and the limit \(E_{\infty}\) can be embedded isometrically in the same metric space \(E\) in a such a way that the convergence holds in the sense of the Hausdorff distance. Section 3 titled ``The construction of the Brownian sphere'' introduces the concept of a snake trajectory (this is a convenient framework for studying the Brownian snake), see \textit{C. Abraham} and the author [J. Eur. Math. Soc. (JEMS) 20, No. 12, 2951--3016 (2018; Zbl 1501.60046)]. Section 4 titled ``Discrete bijections with trees'' presents the simplest example of the bijection between planar maps and labeled trees, in the case of quadrangulations. Section 5 titled ``Infinite-volume models and the Brownian plane''briefly presents the Brownian plane, which is an infinite-volume version of the Brownian sphere and can be obtained as the scaling limit of the infinite random lattices called the UIPT (for uniform infinite planar triangulation) and the UIPQ (for uniform infinite planar quadrangulation). Section 6 titled ``Planar maps with a boundary and Brownian disks'', introduces Brownian disks as scaling limits of planar maps with a boundary, when the boundary size tends to infinity, see [\textit{J. Bettinelli} and \textit{G. Miermont}, Probab. Theory Relat. Fields 167, No. 3--4, 555--614 (2017; Zbl 1373.60062)]. In Section 7 titled ``Excursion theory for Brownian motion indexed by the Brownian tree'' the author proves that excursions are independent conditionally given their ``boundary sizes'', and distributed according to a certain excursion measure on snake trajectories. Section 8 titled ``Constructing Brownian disks from the positive Brownian snake excursion measure'' constructs Brownian disks from a continuous random tree equipped with Brownian labels, which is analogous to the construction of the Brownian sphere, with difference that labels now correspond to distances from the boundary (this is in contrast with the previous constructions in [Zbl 1373.60062]), which also used labeled trees, but with a different interpretation of labels. In particular, Section 8 shows that connected components of the complement of the ball of radius \(r\) centered at the distinguished point in the Brownian sphere are independent Brownian disks, conditionally on their boundary sizes and volumes. Finally, Section 9 titled ``Slicing Brownian disks at heights'' studies the sequence of boundary sizes of the connected components of \(\{x\in D:\, H(x)>r\}\) as a process parametrized by \(r\) (\(D\) is a free Brownian disk and \(H(x)\) denotes the distance from point \(x\in D\) to the boundary). The author shows that this process is a growth-fragmentation process whose distribution is completely determined. The latter result is very closely related to the investigation of the paper [Zbl 1373.60062] sealing limits for a similar process associated with triangulation with a boundary.