Redundancy Channels in the Conformal Bootstrap
Stefanos R. Kousvos, Andreas Stergiou · 2025
Read Redundancy Channels in the Conformal Bootstrap by Stefanos R. Kousvos, Andreas Stergiou (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-05338.txt.
Introduction
Stefanos Kousvos and Andreas Stergiou tackle a snag in a numerical method used to pin down quantum field theories, in a research paper from 2025. The method, called the conformal bootstrap, starts from a chosen symmetry. Sometimes it quietly allows theories with a larger symmetry, which blurs the results.
The authors trace this to certain operators that are redundant in the smaller-symmetry theory. Blocking those operators with a gap can separate look-alike theories. They apply the idea to tell a cubic theory apart from a rounder, more symmetric one in three dimensions. A clean channel of excluded values appears exactly where the symmetry would have grown.
Catalog record
| CREATOR | Stefanos R. Kousvos, Andreas Stergiou |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 74 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-05338
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Redundancy Channels in the Conformal Bootstrap Stefanos R. Kousvos, Andreas Stergiou Abstract A method for obstructing symmetry enhancement in numerical conformal bootstrap calculations is proposed. Symmetry enhancement refers to situations where bootstrap studies initialised with a certain symmetry end up allowing theories with higher symmetry. In such cases, it is shown that redundant operators in the less symmetric theory can descend from primary scaling operators of the more symmetric one, motivating the imposition of spectral gaps that are justified in the former but not the latter. The same mechanism can also be used to differentiate between decoupled and fully coupled theories which otherwise have the same global symmetry. A systematic understanding of this mechanism is developed and applied to distinguish the cubic from the O ( 3 ) O(3) model in three dimensions, where a strip of disallowed parameter space, referred to as the cubic redundancy channel, emerges once a gap associated with a redundant operator of the cubic theory is imposed. The channel corresponds precisely to the region of parameter space where the assumed cubic symmetry would be enhanced to O ( 3 ) O(3) . Contents 1 Introduction The numerical conformal bootstrap [ 1 , 2 ] has given us unprecedented access to strongly-coupled conformal physics. Most notably, it has revolutionised our understanding of the operator spectrum of the 3D Ising model [ 3 , 4 , 5 , 6 , 7 ] , as well as that of O(N) models [ 8 , 9 , 10 , 11 ] . While these successes fully attest to the power and importance of the numerical conformal bootstrap, it is of great value to assess its strength in less symmetric cases, with richer operator spectra. This programme has been hindered by the problem of accidental enhancement of global symmetry, i.e. the problem of bootstrap constraints obtained with some global symmetry in mind allowing a theory with larger global symmetry [ 12 , 13 , 14 , 15 ] . In this work, we provide a methodology to overcome this problem, and demonstrate explicitly how it works for cubic global symmetry in d=3 dimensions. A large part of the perturbative spectrum of this theory is already known from \varepsilon expansion methods [ 16 , 17 , 18 , 19 ] , further motivating a non-perturbative bootstrap study. Our proposal is centred around redundant operators. In a Lagrangian formalism, redundant operators are proportional to the equation of motion and can be removed by field redefinitions. As such, they do not generate genuine physical effects and are modded out of the CFT spectrum. This, however, does not mean that they can be ignored in treatments of renormalisation [ 20 , 21 ] or in discrete spin systems [ 22 ] and Monte Carlo simulations. Indeed, they are known to be crucial for proper accounting of operator mixing and the derivation of anomalous dimensions, as well as the satisfaction of Ward identities. The fact that redundant operators vanish on-shell implies that their correlation functions are contact terms (semi-local or ultra-local). Such correlation functions are of no relevance to the numerical conformal bootstrap, which is based on the constraints of unitarity and crossing symmetry of position-space four-point functions in configurations where all operator insertion points are separated. Despite this, redundant operators have in fact been used in the numerical conformal bootstrap for a long time, albeit through their absence as exchanged operators in relevant operator product expansions (OPEs). For example, the 3D Ising island is obtained assuming a single relevant scalar \mathbb{Z}_{2} -odd operator, commonly denoted by \sigma , exchanged in the \sigma\times\epsilon OPE, where \epsilon is the leading scalar \mathbb{Z}_{2} -even operator (beyond the identity). In a weakly coupled description, the next-to-leading scalar \mathbb{Z}_{2} -odd operator, \sigma^{\prime} , would be given schematically by \phi^{3} , but, as we will review below, this operator becomes redundant. This makes \sigma^{\prime} schematically of the form of \phi^{5} and motivates a large gap between \sigma and \sigma^{\prime} , which ends up being crucial in isolating the 3D Ising model in a bootstrap island. Such perturbatively inspired considerations were given a more formal flavour in [ 23 ] . The main message of this work is that gaps associated with redundant operators are much more widely useful in the numerical conformal bootstrap. For our case of interest, that of unwanted symmetry enhancements from H to G with H a subgroup of G , the crucial observation is that redundant operators in the theory with symmetry H may be drawn from scaling, non-redundant operators of the theory with symmetry G . This motivates the introduction of gaps in the H -symmetric theory that are absent in the corresponding sector of the G -symmetric one, if there are no nearby non-redundant operators with the same quantum numbers in the H -symmetric theory. Such gaps, then, preclude the enhancement of H to G . Equation of motion inspired gaps also allow us to differentiate between decoupled and fully coupled theories, based on the observation that a fully coupled theory has one stress-energy energy tensor, whereas decoupled theories have multiple. While our proposal enjoys a high-degree of generality, which we will outline in various examples arising via the \varepsilon expansion [ 24 , 25 , 26 ] , we will mostly examine it here in the context of a theory with cubic global symmetry in d=3 . A theory with hypercubic global symmetry, C_{N}=\mathbb{Z}_{2}{\!}^{N}\rtimes S_{N} , can be defined in d=4-\varepsilon starting with N scalar fields, \phi_{i} , i=1,\ldots,N , with interactions described by the action S=\int d^{\hskip 1.0pt4-\varepsilon}x\,\big(\tfrac{1}{2}\partial^{\mu}\phi_{i}\hskip 1.0pt\partial_{\mu}\phi_{i}+\tfrac{1}{8}\lambda(\phi^{2})^{2}+\tfrac{1}{4!}g\hskip 1.0pt\delta_{ijkl}\phi_{i}\phi_{j}\phi_{k}\phi_{l}\big)\,, (1.1) where \delta_{ijkl} is the generalised Kronecker delta symbol, which is equal to one when all indices take the same value and zero otherwise. Besides the free theory ( \lambda=g=0 ) and the theory of N decoupled Ising models ( \lambda=0,g\neq 0 ), this action is known to have two fully interacting fixed points, namely the O(N) model ( \lambda\neq 0,g=0 ) and the hypercubic model ( \lambda g\neq 0 ). At large N , it is known that the infrared (IR) stable fixed point is the hypercubic one, while, as N is lowered, the hypercubic fixed point approaches the O(N) one. At some critical value of N , N_{c} , these two fixed points pass through each other. At that point, they exchange their stability properties and below N_{c} it is the O(N) theory that is IR stable [ 27 , 28 , 29 , 24 , 25 ] . This is summarised in Fig. 1 . \lambda g G H C I N>N_{c} \lambda g G H I C N<N_{c} Fig. 1: Depending on the value of N , the IR stable fixed point is either the O(N) model, denoted by H, or the hypercubic theory, denoted by C. The free theory is denoted by G and decoupled Ising models by I. The region between the hatched lines is the basin of attraction of the IR stable fixed point. The value N_{c} is equal to four at leading order in the \varepsilon expansion, but gets corrected order by order in \varepsilon . Early studies can be found in [ 30 , 31 , 32 , 33 ] ,