[Paper Review] Corrected Loop Vertex Expansion for Phi42 Theory

Vincent Rivasseau, Zhituo Wang|arXiv (Cornell University)|Jun 28, 2014

Noncommutative and Quantum Gravity Theories7 references4 citations

TL;DR

This paper corrects a critical error in the cleaning expansion method of the Loop Vertex Expansion (LVE) for the $φ^4_2$ quantum field theory, replacing it with a multiscale LVE that preserves renormalization constraints across slices. The authors establish Borel summability of the free energy in a cardioid domain with opening angle arbitrarily close to $2\pi$, exceeding standard constructive bounds, via improved convergence from marked propagator scaling and combinatorial control.

ABSTRACT

This paper is an extended erratum to J. Math. Phys.53, 042302 (2012) and arXiv:1104.3443, in which the classic construction and Borel summability of the phi^4_2 Euclidean quantum field theory was revisited combining a multi-scale analysis with the constructive method called Loop Vertex Expansion (LVE). Unfortunately we discovered an important error in the method of J. Math. Phys.53, 042302 (2012). We explain the mistake, and provide a new, correct construction of the phi^4_2 theory according to the LVE.

Motivation & Objective

• To correct a fundamental error in the cleaning expansion method of the Loop Vertex Expansion (LVE) for $φ^4_2$ theory, which failed to converge due to unaccounted scale sums in tadpole contributions.
• To develop a new, rigorous construction of the $φ^4_2$ theory using a corrected multiscale LVE that maintains hardcore renormalization constraints across slice levels.
• To prove analyticity of the free energy in a coupling constant domain with opening angle arbitrarily close to $2\pi$, larger than standard Nevanlinna-Sokal or Watson-type domains.
• To lay the groundwork for applying the corrected multislice LVE to just-renormalizable and non-local quantum field theories, such as non-commutative or tensor group field theories.

Proposed method

• Replaces the flawed cleaning expansion with a multislice Loop Vertex Expansion (MLVE) that enforces constraints between exponential interactions across different scale slices.
• Uses a slice-wise decomposition of the propagator $C = \sum_{j=0}^{j_{\text{max}}} C_j$, with $C_j$ corresponding to momentum shell $M^{-2j}$ to $M^{-2(j-1)}$, and introduces a multiscale parametric representation.
• Applies a forest formula with recursive cuts to generate families of graphs $F^\mathcal{C}_m(G)$, ensuring each initial c-propagator is copied into $2^m$ copies across the family with identical slice attribution.
• Implements a geometric average bound on amplitudes via Lemma 4.6, replacing $|\lambda|$ with $\rho$ such that $|\lambda| \cos^{-2}(\phi/2) \leq \rho$ in the cardioid domain.
• Binds the amplitude of each graph $G'$ using $|A_{G'}| \leq O(1)^n \rho^n p^{8n} M^{-2\sum_{i=1}^p j_i / 7}$, where $p$ is the number of marked non-tadpole propagators and $j_i$ their scales.
• Controls combinatorial factors via bounds on partitions, trees, and derivative actions, all dominated by the $M^{-O(1)p^2}$ convergence factor from marked propagators.

Experimental results

Research questions

• RQ1 How can the flawed cleaning expansion in the LVE for $φ^4_2$ theory be corrected to restore convergence under renormalization?
• RQ2 Can a multiscale LVE be constructed such that renormalization constraints are preserved across slice levels, preventing uncontrolled counterterm proliferation?
• RQ3 What is the maximal domain of analyticity for the free energy in the coupling constant $\lambda$ when using the corrected MLVE method?
• RQ4 Can the MLVE technique be extended to treat just-renormalizable or non-local quantum field theories beyond the $φ^4_2$ model?

Key findings

• The corrected multislice LVE successfully constructs the $φ^4_2$ theory with ultraviolet cutoff removed, restoring convergence lost in the original cleaning expansion.
• The free energy is proven to be analytic in a cardioid-shaped domain in the complex $\lambda$-plane with opening angle arbitrarily close to $2\pi$, exceeding standard constructive bounds.
• The convergence of the series is achieved via a $M^{-O(1)p^2}$ factor from marked non-tadpole propagators, which dominates all combinatorial and auxiliary expansion factors.
• The method bounds all amplitudes uniformly using a geometric average over graph families, ensuring convergence independent of the initial graph structure.
• The corrected approach avoids the flaw in the original paper by properly accounting for scale sums over tadpoles, which were previously omitted in section 5.
• The framework is generalizable to more complex models, including non-commutative and tensor group field theories, due to its robust multiscale and combinatorial structure.