[Paper Review] Dual Non-Abelian Yang-Mills Simulations in Four Dimensions

J. Wade Cherrington|ArXiv.org|Oct 10, 2009

Quantum Chromodynamics and Particle Interactions25 references3 citations

TL;DR

This paper presents the first Metropolis Monte Carlo simulation of four-dimensional $SU(2)$ Yang-Mills theory using dual variables based on spin foam duality. By employing an $O(j^4)$ algorithm for the 48-spin vertex amplitude and a novel dual Metropolis algorithm with homology, cube, and edge moves, the authors validate the dual approach against conventional simulations, demonstrating feasibility despite the severe sign problem at low $eta$ values.

ABSTRACT

We present numerical results for pure SU(2) Yang-Mills theory in four space-time dimensions using a novel algorithm based on dually transformed variables. The simulation makes use of a recently derived O(j^4) algorithm for the dual vertex amplitude and a dual Metropolis algorithm that generalizes the one recently developed for three dimensions. The dual algorithm is validated against the equivalent model using conventional variables over a range of couplings, spin cut-offs, and lattice sizes. We consider a lattice size up to 8x8x8x8, where the problem of negative amplitudes renders the simulation results excessively noisy even at a relatively low beta (starting at about beta=1.8). In conclusion, we survey some approaches to addressing the sign problem in this context and increasing the efficiency of dual computations within this approach.

Motivation & Objective

• To develop and validate a dual algorithm for pure $SU(2)$ Yang-Mills theory in four dimensions using duality transformations.
• To address the challenge of constructing ergodic moves and computable amplitudes in non-Abelian dual lattice gauge theories.
• To test the performance and reliability of the dual algorithm across varying couplings, lattice sizes, and spin cutoffs.
• To assess the severity of the sign problem in the dual formulation, particularly at low $eta$ values.
• To lay the groundwork for extending dual simulations to include observables like Wilson and Polyakov loops.

Proposed method

• The dual model uses irreducible representations ($j$) on plaquettes and intertwiners ($i^r$) on edges, with admissibility constraints enforced via triangle inequalities.
• The vertex amplitude is computed using an $O(j^4)$ algorithm derived in [CC2009], enabling efficient evaluation of the 48-spin amplitude in 4D hypercubic lattices.
• A dual Metropolis algorithm is implemented using three move types: homology moves (global spin shifts on closed planes), cube moves (local plaquette spin changes), and edge moves (relative intertwiner shifts by ±2 units).
• The intertwiner labels are stored as offsets from the minimum admissible value, allowing efficient computation and dynamic range handling.
• The acceptance ratio is determined by the local amplitude ratio before and after each move, following the standard Metropolis criterion.
• The algorithm is validated by comparing expectation values of an effective observable against results from conventional lattice simulations across multiple lattice sizes and couplings.

Experimental results

Research questions

• RQ1 Can a dual algorithm based on spin foam duality be successfully implemented for $SU(2)$ Yang-Mills theory in four dimensions?
• RQ2 How does the performance of the dual algorithm compare to conventional simulations in terms of accuracy and efficiency?
• RQ3 To what extent does the sign problem affect the dual simulation, particularly at low $eta$ values?
• RQ4 Are the proposed ergodic moves—homology, cube, and edge—sufficient to ensure full exploration of the configuration space?
• RQ5 Can the $O(j^4)$ vertex amplitude algorithm be effectively integrated into a Markov chain Monte Carlo framework for non-Abelian gauge theories?

Key findings

• The dual algorithm successfully reproduces conventional simulation results across a range of couplings, spin cutoffs, and lattice sizes, validating its correctness.
• The simulation was performed on lattice sizes up to $8^4$, demonstrating scalability of the dual approach in four dimensions.
• The sign problem becomes severe at low $eta$, with negative amplitudes causing excessive noise starting around $eta = 1.8$.
• The use of relative intertwiner labels ($i^r$) significantly improves computational efficiency and simplifies move implementation.
• The cube move mechanism implicitly adjusts intertwiners through admissibility range changes, reducing the need for explicit intertwiner tracking.
• The algorithm remains ergodic in practice, as configurations with zero amplitude do not appear to disconnect the configuration space.