The hypergraph stochastic block model is a statistical model for sampling inhomogeneous random hypergraphs associated with a partition of the vertex set. In this talk I will discuss fundamental concepts and recent developments on the statistical analysis of hypergraph stochastic block models. The focus is on universal information-theoretic bounds and phase transitions that help to understand requirements on data sparsity and model dimensions under which consistent learning of the underlying vertex partition is possible.
I will discuss the basic tools of discrete complex analysis on Z 2 that we use to construct a Virasoro representation on the space of local fields of the discrete Gaussian Free Field. I will also go over some recent results we obtained with Delara Behzad and Kalle Kytölä.
In this expository talk, I will give an outline of the developments in the field of critical and massive scaling limits in the Ising model in two dimensions starting from the breakthrough work of Smirnov, which defined the notion of s-holomorphicity. Emphasis will be placed on explaining the steps and techniques used to relate this discrete complex analytic notion to analysis in continuum, yielding conformal invariance of the model in the critical case.
The correlation functions in conformal field theories, in particular, in minimal models, enjoy a number of properties. Among them are, in particular, Belavin-Polyakov-Zamolodchikov equations, which are second order partial differential equations, and the Operator product expansion (OPE) hypothesis concerning the asymptotic expansions of correlations to all orders. The correlations in scaling limit of the critical the Ising model have been recently computed rigorously. However, the program of relating them on mathematical level to correlations in a minimal CFT is not completed, and deriving the BPZ equations and the OPE from the explicit expressions is not straightforward. In this talk, I will discuss our recent results in this direction. This is a joint work with Christian Webb.
The free fermion was one of Segal's original examples of a theory satisfying his Conformal Field Theory (CFT) axioms. This was made fully rigorous relatively recently by James Tener. I will review Segal's definition of a CFT, and then describe the free fermion. Then, I will report on some of my ongoing work attempting to extend the free fermion, following Stolz and Teichner.
Introduced by Polyakov in the 1980s, Liouville quantum gravity (LQG) is in some sense the canonical model of a random fractal Riemannian surface, constructed using the Gaussian free field. Sheffield showed that when a certain type of LQG surface, called a quantum wedge, is decorated by an appropriate independent SLE curve, the wedge is cut into two independent surfaces which are themselves quantum wedges, and that these resulting wedges uniquely determine the original surface as well as the SLE interface. We prove that the original surface can in fact be obtained as a metric space quotient of the LQG metrics on the two wedges. This was proven by Gwynne and Miller in the special case $\gamma = \sqrt{8/3}$, for which $\gamma$-LQG surfaces are equivalent to Brownian surfaces, allowing an explicit description of the metric in terms of Brownian motion that is not available in general. I will explain how our work uses GFF techniques to extend their results to the whole subcritical regime $\gamma \in (0,2)$, while establishing new estimates describing the boundary behaviour of LQG. Joint work with Jason Miller.
The possible asymptotic distributions of a random dynamical system are described by stationary measures, and in this talk we will be interested in the properties of these measures - in particular, whether they are absolutely continuous. First, I will quickly describe the case of Bernoulli convolutions, which can be seen as generalisations of the Cantor middle third set, and then the case of random iterations of matrices in SL(2, R) acting on the real projective line, where the stationary measure is unique under certain conditions, and is called the Furstenberg measure. It had been conjectured that the Furstenberg measure is always singular when the random walk has a finite support. There have been several counter-examples, and the aim of the talk will be to describe that of Bourgain, where the measure even has a very regular density. I will explain why the construction works for any simple Lie group, using the work of Boutonnet, Ioana, and Salehi Golsefidy on local spectral gaps in simple Lie groups.
The double dimer model (DDM) on a planar graph is a model of random loops, and the Gaussian free field (GFF) is a model of a height function. The two models are linked by a conjecture that the DDM loops converge in the scaling limit to loops in the continuum, which are level lines of the GFF. I will introduce these two models and outline two results. First, certain 2n-point correlation functions in the DDM are known to be determinants in the 2-point functions; we give a new proof of this, which in particular we show can be extended to the GFF. Second, it is known that the GFF exhibits a "height gap": if one sets the boundary height to be +\lambda on one half of the boundary, and -\lambda on the other, then if \lambda = \sqrt(pi/8), the zero level line actually exhibits a sharp "jump" of 2\lambda. We give a simple derivation of this special value of the height gap, using the determinental result. Joint work with Marcin Lis (TU Vienna)
In the late 80's, the physicists Dotsenko and Fateev used the Coulomb-gas formalism to solve for the structure constants of Belavin--Polyakov--Zamolodchikov's minimal models of 2D CFT. To this day, their analysis has not been made mathematically rigorous. In this talk, we will discuss progress towards a rigorous Coulomb-gas formalism, along with its application to the construction of the minimal models.
In the late 80's, the physicists Dotsenko and Fateev used the Coulomb-gas formalism to solve for the structure constants of Belavin--Polyakov--Zamolodchikov's minimal models of 2D CFT. To this day, their analysis has not been made mathematically rigorous. In this talk, we will discuss progress towards a rigorous Coulomb-gas formalism, along with its application to the construction of the minimal models.
In recent decades, there have been interesting connections made between a number of areas of mathematics, including statistical mechanics, knot theory, quantum groups, and subfactors. The Temperley-Lieb algebras are a family of finite dimensional algebras that were the original source of these connections. In this talk we review the representation theory of the finite Temperley-Lieb algebras. We then discuss extremal traces, their classification, and applications to the representation theory of an infinite dimensional generalization of the Temperley-Lieb algebra.
Conformal nets provide a rigorous mathematical framework for conformal field theory, assigning an algebra of observables to each region of the underlying spacetime manifold. Here, we consider so-called discrete conformal nets whereby the spacetime is not a smooth manifold but instead has an 'atomic' structure. It turns out that planar algebras can be used to construct 'almost' examples of discrete conformal nets. In this talk, I will review this business and detail recent efforts to construct fully-fledged examples of discrete conformal nets. Inspired by the statistical mechanics literature, I will also introduce some integrable operators that act on the spacetime and detail some of their algebraic structure.
Log-correlated Gaussian fields such as the Gaussian free field are random Schwartz distributions whose covariances have a logarithmic singularity on their diagonal. They appear for instance in Liouville quantum gravity, characteristic polynomials of random matrices, the dimer model etc. In this talk I will present ways to decompose general log-correlated fields into a sum of a canonical log-correlated field with a particularly nice covariance structure and a Hölder-continuous error term. I will also discuss applications to the study of the Gaussian multiplicative chaos of the field. The talk is based on joint works with Eero Saksman and Christian Webb as well as Juhan Aru and Antoine Jego.
I discuss the recent preprint [arXiv:2211.16111] of Nikolay Barashkov and I, where we show the almost sure well-posedness of a deterministic nonlinear wave equation (cubic Klein-Gordon equation) on the plane. Here "almost sur" is in respect to the \\\\phi^4 quantum field theory. I briefly introduce the invariant measure argument and outline the solution on 2D torus due to Oh and Thomann. I then explain our main contributions: extension of periodic solutions to infinite volume, and a weaker result for nonlinear Schrödinger equation. The viewpoint is functional-analytic with a dash of probability.
One method of introducing external randomness to a Gibbs state, as opposed to the internal randomness of the Gibbs state itself, is to perturb the Hamiltonian with a term corresponding to the coupling of a random external field to the system. For the mean-field spherical model, the corresponding perturbed model can be exactly solved, in some sense, in the infinite volume limit. In this talk, we will introduce, motivate, and present some constructions and results concerning the so-called infinite volume metastases of the mean-field spherical model in a random external field. The aim of this talk is to present the general theory of disordered systems as it pertains to this particular model, and highlight the particular aspects of this model which lead to its curious behaviour as a disordered system. This talk is based on work in a recently accepted paper to appear in the Journal of Statistical Physics.
I will discuss recent approaches to characterising the Gaussian free field in the plane, and in higher dimensions. The talk will be based on joint work with Juhan Aru, Nathanael Berestycki, and Gourab Ray.
The forward evolution of chaotic systems notoriously washes out inexact information about their state. When advected by a chaotic system, physically relevant measures therefore often converge to some reference measure, usually the SRB measures. This property implies various important statistical behaviours of chaotic systems. In this talk we discuss the behaviour of slices of these physical measures along smooth submanifolds that are reasonably generic (e.g. not stable or unstable manifolds). We give evidence that such conditional measures also have exponential convergence back to the full SRB measures, even though they lack the regularity usually required for this to occur (for example, they may be Cantor measures). Using Fourier dimension results, we will prove that CDoC holds in a class of generalised baker's maps, and we will give rigorous numerical evidence in its favour for some non-Markovian piecewise hyperbolic maps. CDoC naturally encodes the idea of long-term forecasting of systems using perfect partial observations, and appears key to a rigorous understanding of the emergence of linear response in high-dimensional systems.
We consider a random height function associated with the dimer model on a graph embedded into a Riemann surface. Given a sequence of such graphs approximating the surface in a certain sense we prove that the corresponding sequence of height functions converges to the compactified free field on the surface. To establish this result we follow approach developed by Dubédat: we introduce a family of observables of the model which can be expressed as determinants of discrete perturbed Cauchy-Riemann operators, we analyze the latter using Quillen curvature formula.
Page content by: webmaster-math [at] list [dot] aalto [dot] fi