A simple topological number from monopole loops helps locate the deconfinement transition in SU(3) lattice theory
This paper introduces a new, simple observable called “simplicity” that summarises the loop structure of magnetic monopole currents in lattice simulations of SU(3) Yang–Mills theory. Simplicity is defined from two basic topological counts of the monopole network: b0, the number of connected pieces of the network, and b1, the number of independent loops. The authors show that measuring the average of the ratio b0/b1 across gauge field configurations gives a sharp signal of the temperature where confinement is lost.
To build this observable the authors work on a standard four‑dimensional lattice with the Wilson action. They fix the Maximal Abelian Gauge (a partial gauge choice that makes Abelian monopoles easy to find), extract the Abelian monopole currents on the dual lattice, and view those currents as a graph. From that graph they compute b0 and b1 for each configuration and form the simplicity λ = b0/b1. They then study the expectation value of λ and its susceptibility as they change the coupling β, which in lattice practice controls the temperature (the temperature is inversely proportional to the temporal lattice size Nt and the lattice spacing a(β)). The paper reports numerical scans at Nt = 4, 6 and 8 and several spatial sizes, with hundreds of configurations sampled at many β values.
The physical idea is simple. In the low‑temperature, confined phase the monopole network tends to form one large, percolating component with many loops. That gives b0 small and b1 large, so λ is close to zero. At high temperature, monopole currents appear as a sparse gas of small loops, so b0 and b1 are similar and λ approaches one. Across the transition λ rises from near zero toward one. The authors use the peak of the susceptibility of λ and finite‑size scaling in the spatial volume to extract the critical coupling. They report that this topological observable locates the deconfinement point with smaller statistical errors than conventional measures at comparable computational cost.