1-loop renormalisability of integrable sigma-models from 4d Chern-Simons theory

TL;DR

This work establishes that a broad class of classically integrable 2d sigma-models, constructed as surface defects in 4d Chern-Simons theory, are renormalisable at 1-loop. The authors derive a universal expression for the 1-loop divergences in terms of the Lax connection, recast it as a 4d counterterm, and show that the divergences can be absorbed by a flow of the twist 1-form $\omega$ via $\frac{d}{dt}\omega = \partial_z\Psi(z)$. They demonstrate the flow also governs boundary data (and, for elliptic models, torus moduli), with detailed matching of distributional and defect terms ensuring consistency across rational and elliptic cases. The methods unify rational and elliptic integrable sigma-models under a single renormalisation framework and illuminate the geometric content of the RG flow within the 4d Chern-Simons construction. These results provide a robust, general approach to the quantum stability of integrable worldsheet theories and point toward further extensions to more general defect and genus settings.

Abstract

Large families of integrable 2d sigma-models have been constructed at the classical level, partly motivated by the utility of integrability on the string worldsheet. It is natural to ask whether these theories are renormalisable at the quantum level, and whether they define quantum integrable field theories. By considering examples, a folk theorem has emerged: the classically integrable sigma-models always turn out to be renormalisable, at least at 1-loop order. We prove this theorem for a large class of models engineered on surface defects in the 4d Chern-Simons theory by Costello and Yamazaki. We derive the flow of the 'twist 1-form' (a 4d coupling constant that distinguishes different 2d models), proving earlier conjectures and extending previous results. Our approach is general, using the 'universal' form of 2d integrable models' UV divergences in terms of their Lax connection and reinterpreting the result in the language of 4d Chern-Simons. These results apply equally to rational, trigonometric and elliptic models.

Figures (4)

• Figure 1: A cartoon of the defining data of the 4d Chern-Simons theory on $\Sigma\times \mathbb{CP}^1$. The vertical ($x$) direction depicts the 2d manifold $\Sigma$, while the horizontal ($z$) direction is the spectral $\mathbb{CP}^1$. The meromorphic twist 1-form $\omega$ is defined by its poles (solid lines) and zeros (dashed lines). These marked points in the spectral variable define 2d defects where certain boundary conditions must be imposed. The zeros of $\omega$ are so-called 'disorder defects', where the 4d gauge field may become singular; they split into two subsets $\widehat{\mathbb{Z}}^+$ (in red) and $\widehat{\mathbb{Z}}^-$ (in blue) where the light-cone components $A_+$ and $A_-$ may diverge. At the poles of $\omega$, where the 2d theory lives, the boundary condition is a choice of maximally isotropic subalgebra $\mathfrak{b}$ of the defect algebra $\mathfrak{d}$, where the jet of the 4d gauge field must lie.
• Figure 2: The Feynman rules corresponding to the action in the exponent of the path integral \ref{['PI']}. Solid and dashed lines denote fluctuation and background fields, respectively. The fields are algebra-valued and decomposed as $u=u^\alpha\, T_\alpha$, where Greek indices run over a basis of algebra generators $\left\{T_\alpha\right\}_{\alpha=1}^{\text{dim}\mathfrak{g}}$ satisfying $[T_\beta, T_\gamma] = f^\alpha{}_{\beta\gamma} T_\alpha$ and are raised and lowered with $\kappa_{\alpha \beta}\equiv \langle T_\alpha, T_\beta \rangle$ and its inverse. The light-cone components of the 2d momenta are denoted by $k_\pm$.
• Figure 3: In the 'universal' formulation, the divergent part of the 1-loop effective action \ref{['ren']} is given by a single Feynman diagram.
• Figure 4: The original unit cell representing $\mathbb{T}$ (marked in white) with basis vectors $\lambda_1$ and $\lambda_2$ and boundary marked in black. Under RG, the surface changes its shape and is defined in terms of new basis vectors $\lambda_1'$ and $\lambda_2'$: this introduces an additional area (marked in red) that should be integrated over in the effective action, which in turn results in $C_{\text{moduli}}$. $\mathcal{C}_1$ and $\mathcal{C}_2$ denote the contours along the left and bottom boundaries, respectively, with orientation indicated by the arrows.

Theorems & Definitions (2)

• proof
• proof