One-loop beta-functions of quartic enhanced tensor field theories

TL;DR

This paper computes one-loop beta-functions for two quartic enhanced tensor field theories, revealing their renormalization group behaviors and fixed points, which differ from conventional tensor models and may avoid branched polymer phases.

Contribution

It introduces and analyzes two new quartic enhanced tensor field theories, providing their one-loop beta-functions and exploring their asymptotic safety and renormalization group flows.

Findings

Model + exhibits asymptotic safety with a fixed point for one coupling.

Model × has a nontrivial RG flow with some couplings vanishing in the UV.

Both models have constant wave function renormalization, no anomalous dimension.

Abstract

Enhanced tensor field theories (eTFT) have dominant graphs that differ from the melonic diagrams of conventional tensor field theories. They therefore describe pertinent candidates to escape the so-called branched polymer phase, the universal geometry found for tensor models. For generic order $d$ of the tensor field, we compute the perturbative $\beta$-functions at one-loop of two just-renormalizable quartic eTFT coined by $+$ or $\times$, depending on their vertex weights. The models $+$ has two quartic coupling constants $(\lambda, \lambda_{+})$, and two 2-point couplings(mass, $Z_a$). Meanwhile, the model $\times$ has two quartic coupling constants $(\lambda, \lambda_{\times})$ and three 2-point couplings (mass, $Z_a$, $Z_{2a}$). At all orders, both models have a constant wave function renormalization: $Z=1$ and therefore no anomalous dimension. Despite such peculiar behavior, both models acquire nontrivial radiative corrections for the coupling constants. The RG flow of the model $+$ exhibits a particular asymptotic safety: $\lambda_{+}$ is marginal without corrections thus is a fixed point of arbitrary constant value. All remaining couplings determine relevant directions and get suppressed in the UV. Concerning the model $\times$, $\lambda_{\times}$ is marginal and again a fixed point (arbitrary constant value), $\lambda$, $\mu$ and $Z_a$ are all relevant couplings and flow to 0. Meanwhile $Z_{2a}$ is a marginal coupling and becomes a linear function of the time scale. This model can neither be called asymptotically safe or free.