The Large N Random Phase sine-Gordon Model

TL;DR

This paper introduces a large N version of the 2D random phase sine-Gordon model and uses advanced techniques to show that the correlation functions have a specific logarithmic form with a suppressed coefficient B, clarifying discrepancies in previous predictions.

Contribution

It develops a large N generalization of the model and demonstrates through non-Abelian bosonization and RG that the correlation function coefficient B is suppressed by 1/N^3, resolving conflicting predictions.

Findings

Correlation functions follow the predicted logarithmic form.

Coefficient B is suppressed by 1/N^3 compared to A.

Provides a theoretical framework for large N analysis of the model.

Abstract

At large distances and in the low temperature phase, the quenched correlation functions in the 2d random phase sine-Gordon model have been argued to be of the form~: $ \bar {\vev{~[\varphi(x)-\varphi(0)]^2~}}_* = A (\log|x|) + B \ep^2 (\log|x|)^2 $, with $\ep=(T-T_c)$. However, renormalization group computations predict $B\not=0$ while variational approaches (which are supposed to be exact for models with a large number of components) give $B=0$. We introduce a large $N$ version of the random phase sine-Gordon model. Using non-Abelian bosonization and renormalization group techniques, we show that the correlation functions of our models have the above form but with a coefficient $B$ suppressed by a factor $1/N^3$ compared to $A$.