Towers of Operators in CFTs and Convexity Bounds at Large Charge

Technical Summary: Towers of Operators in CFTs and Convexity Bounds at Large Charge

Problem Statement
This paper investigates the scaling dimensions of operators in Conformal Field Theories (CFTs) carrying large global charges. Specifically, it focuses on 3d CFTs with moduli spaces where a continuous global symmetry is spontaneously broken. In such theories, the minimum scaling dimension $\Delta_{\min}(Q)$ in a sector of charge $Q$ follows a large-charge expansion:
$$\Delta_{\min}(Q) = \alpha_1 Q + \alpha_0 + O(1/Q)$$
The authors aim to determine whether the coefficients $\alpha_1$ and $\alpha_0$ are subject to universal bounds motivated by the holographic swampland program, particularly the Weak Gravity Conjecture (WGC). While the WGC suggests a "convexity" condition for charged operators, the precise constraints on the subleading coefficient $\alpha_0$ and the leading slope $\alpha_1$ in general CFTs (especially those with less than four supercharges) remain to be rigorously established.

Methodology
The paper employs two primary methodological approaches:

1. Moduli-Space Effective Field Theory (EFT) and Large-Charge Expansion: The authors utilize the state-operator correspondence to map the problem to a cylinder geometry ($\mathbb{R} \times S^2$). They expand around a homogeneous, helical saddle point representing the large-charge state. The leading coefficient $\alpha_1$ is derived from the classical saddle energy, while $\alpha_0$ arises from the one-loop Casimir energy of quadratic fluctuations around this saddle.
2. Explicit Computations in Double-Scaling Limits: To test general bounds and compute specific values, the authors analyze several 3d $\mathcal{N}=1$ theories using:
   • The $\epsilon$-expansion near $d=4$ dimensions.
   • Large-$N_f$ and large-$N$ limits.
   • Specific models include $SO(N) \times SO(N)$, $SO(3)$ Wess-Zumino models, $SQED$, and $SU(2)$ SQCD.
3. Variational Proofs: For the general proof of bounds, the authors analyze the quadratic fluctuation Hamiltonian of the EFT. They utilize variational principles to compare the true ground-state energy against a reference Hamiltonian to establish the sign of the Casimir energy contribution.

Key Contributions and Results

1. Computations of $\alpha_1$ and $\alpha_0$
The authors provide the first direct computations of the subleading coefficient $\alpha_0$ in several 3d $\mathcal{N}=1$ theories (rather than relying solely on $\epsilon$-expansion continuation).

• $SO(N) \times SO(N)$ Model: In the large-$N$ limit, $\alpha_1 = \sqrt{(Q_x^2 + Q_y^2)/2}$ and $\alpha_0$ is expressed in terms of a regulated determinant function $\xi(x)$, which is found to be non-positive.
• $SO(3)$ Model: $\alpha_1 = 1 - \epsilon/2$ and $\alpha_0 = -1/16$.
• $SQED$ (Large $N_f$): The leading order result yields $\alpha_0 = 0$, consistent with the free-theory limit where the Casimir energy vanishes.
• $SU(2)$ SQCD: The authors compute the scalar and fermion contributions to $\alpha_0$ in terms of a Wilson coefficient $c$. They find that for the large-$N_f$ limit ($c \approx 1$), the fermion contribution vanishes, and the scalar contribution is negative ($\alpha_0 \approx -0.556$).

2. Proof of Projected Convexity Bound ($\alpha_0 \leq 0$)
Motivated by the CFT Convexity Conjecture (CCC), which requires $\Delta_{\min}(Q_1 + Q_2) \geq \Delta_{\min}(Q_1) + \Delta_{\min}(Q_2)$, the asymptotic form implies $\alpha_0 \leq 0$.

• Projected Observable: The authors define a "projected" minimum dimension $\Delta^{(n)}_{\min}(Q)$ where only the charge along a specific Cartan generator $T_n$ is fixed, while charges in orthogonal directions are minimized over.
• Main Theorem: They prove that for this projected observable, the one-loop coefficient satisfies $\alpha^{(n)}_0 \leq 0$.
  • The proof relies on showing that for the saddle minimizing the projected dimension, the mass matrix $M$ of the transverse fluctuations is positive semi-definite ($M \geq 0$).
  • Using a variational argument, they demonstrate that if $M \geq 0$, the scalar contribution to the Casimir energy is non-positive. Combined with known non-positive contributions from fermions and gauge fields, this establishes the bound.
• Limitation: The authors explicitly show that this proof does not extend to the "fixed-charge-ray" problem (where the full charge vector is fixed). They construct a two-derivative EFT counterexample where a fixed charge ray yields $\alpha_0 > 0$, indicating that the stronger convexity statement cannot be proven using moduli-space EFT alone without additional UV input.

3. Absence of Universal Bounds on $\alpha_1$
The paper argues that there is no universal bound on the leading coefficient $\alpha_1$ (other than the trivial $\alpha_1 \geq 0$).

• Sensitivity to Charge Normalization: The value of $\alpha_1$ is shown to be highly sensitive to the choice of the charge lattice and the normalization of the $U(1)$ generator.
• Counterexamples:
  • Gauging a finite subgroup $\mathbb{Z}_N \subset U(1)$ can arbitrarily increase $\alpha_1$ by a factor of $N$.
  • Choosing a primitive direction in a multi-$U(1)$ lattice (e.g., $Q = aQ_x + bQ_y$ with large $a, b$) can arbitrarily decrease $\alpha_1$.
• Normalized Slope: The authors consider a normalized slope $\hat{\alpha}_1 = \alpha_1 \sqrt{C_J/C_T}$ (involving current and stress-tensor two-point functions) to mitigate normalization issues. While this removes the discrete lattice effects, they find no evidence for a universal upper bound on $\hat{\alpha}_1$ either, and note that this normalization obscures the direct geometric interpretation of scale separation.

Significance
The paper claims to provide a concrete, provable CFT bound motivated by the holographic swampland program. Specifically, it establishes that the projected scaling dimension of large-charge towers in 3d CFTs with moduli spaces must satisfy $\alpha_0 \leq 0$. This result is significant because it is derived directly using CFT tools (large-charge EFT and variational methods) rather than relying on holographic duality assumptions.

However, the authors are modest regarding the scope of this result. They emphasize that the bound applies to the projected observable, not necessarily to every fixed-charge ray, and that the stronger fixed-ray convexity conjecture remains unproven and potentially false in general EFTs. Furthermore, the paper clarifies that the leading slope $\alpha_1$ is not a robust diagnostic for scale separation in the swampland sense due to its dependence on discrete charge lattice choices. The work thus delineates the precise limits of what can be proven about large-charge towers using current EFT methods.