Groupe de renormalisation | EPFL Graph Search

En physique statistique, le groupe de renormalisation est un ensemble de transformations qui permettent de transformer un hamiltonien en un autre hamiltonien par élimination de degrés de liberté tout en laissant la fonction de partition invariante. Il s'agit plus exactement d'un semi-groupe, les transformations n'étant pas inversibles. Le groupe de renormalisation permet de calculer les exposants critiques d'une transition de phase. Il permet aussi de prédire la transition Berezinsky-Kosterlitz-Thouless. En physique de la matière condensée, il permet de traiter l'effet Kondo et le liquide de Luttinger. Il a également des applications dans la théorie des systèmes désordonnés. Historique En mécanique statistique, le groupe de renormalisation a été introduit par Kenneth G. Wilson au début des années 1970. Auparavant, en théorie quantique des champs, il avait été étudié comme une invariance de la théorie des champs renormalisée sous l'effet d'une variation des paramètres nus

RG-Flow: a hierarchical and explainable flow model based on renormalization group and sparse prior

Dian Wu

Flow-based generative models have become an important class of unsupervised learning approaches. In this work, we incorporate the key ideas of renormalization group (RG) and sparse prior distribution to design a hierarchical flow-based generative model, RG-Flow, which can separate information at different scales of images and extract disentangled representations at each scale. We demonstrate our method on synthetic multi-scale image datasets and the CelebA dataset, showing that the disentangled representations enable semantic manipulation and style mixing of the images at different scales. To visualize the latent representations, we introduce receptive fields for flow-based models and show that the receptive fields of RG-Flow are similar to those of convolutional neural networks. In addition, we replace the widely adopted isotropic Gaussian prior distribution by the sparse Laplacian distribution to further enhance the disentanglement of representations. From a theoretical perspective, our proposed method has O(log L) complexity for inpainting of an image with edge length L, compared to previous generative models with O(L-2) complexity.

IOP Publishing Ltd2022

Numerical investigations of 3D CFTs using conformal bootstrap methods

Marten Jan Reehorst

Conformal field theories (CFTs) play a very significant role in modern physics, appearing in such diverse fields as particle physics, condensed matter and statistical physics and in quantum gravity both as the string worldsheet theory and through the AdS/CFT correspondence. In recent years major breakthroughs have been made in solving these CFTs through a method called numerical conformal bootstrap. This method uses consistency conditions on the CFT data in order to find and constrain conformal field theories and obtain precise measurements of physical observables. In this thesis we apply the conformal bootstrap to study among others the O(2)- and the ARP^3- models in 3D. In the first chapter we extend the conventional scalar numerical conformal bootstrap to a mixed system of correlators involving a scalar field charged under a global U(1) symmetry and the associated conserved spin-1 current J. The inclusion of a conserved spinning operator is an important advance in the numerical bootstrap program. Using numerical bootstrap techniques we obtain bounds on new observables not accessible in the usual scalar bootstrap. Concentrating on the O(2) model we extract rigorous bounds on the three-point function coefficient of two currents and the unique relevant scalar singlet, as well as those of two currents and the stress tensor. Using these results, and comparing with a quantum Monte Carlo simulation of the O(2) model conductivity, we give estimates of the thermal one-point function of the relevant singlet and the stress tensor. We also obtain new bounds on operators in various sectors. In the second chapter we investigate the existence of a second-order phase transition in the ARP^3 model. This model has a global O(4) symmetry and a discrete Z_2 gauge symmetry. It was shown by a perturbative renormalization group analysis that its Landau-Ginzburg-Wilson effective description does not have any stable fixed point, thus disallowing a second-order phase transition. However, it was also shown that lattice simulations contradict this, finding strong evidence for the existence of a second-order phase transition. In this chapter we apply conformal bootstrap methods to the correlator of four scalars t transforming in the traceless symmetric representation of O(4) in order to investigate the existence of this second order phase transition. We find various features that stand out in the region predicted by the lattice data. Moreover, under reasonable assumptions a candidate island can be isolated. We also apply a mixed t-s bootstrap setup in which this island persists. In addition we study the kink-landscape for all representations appearing in the t times t OPE for general N. Among others, we find a new family of kinks in the upper-bound on the dimension of the first scalar operator in the "Box" and "Hook" representations.

EPFL2020

Constraints on perturbative RG flows in six dimensions

Lorenzo Vitale

When conformal field theories (CFTs) are perturbed by marginally relevant deformations, renormalization group (RG) flows ensue that can be studied with perturbative methods, at least as long as they remain close to the original CFT. In this work we study such RG flows in the vicinity of six-dimensional unitary CFTs. Neglecting effects of scalar operators of dimension two and four, we use Weyl consistency conditions to prove the alpha-theorem in perturbation theory, and establish that scale implies conformal invariance. We identify a quantity that monotonically decreases in the flow to the infrared due to unitarity, showing that it does not agree with the one studied recently in the literature on the six-dimensional O-3 theory.

Springer2016

Numerical investigations of 3D CFTs using conformal bootstrap methods

Marten Jan Reehorst

EPFL2020