Compact U(1) Lattice Gauge Theory — 2+1D Monte Carlo

pure gaugeWilson actionMetropolis + over-relaxation
mode
idle
sweep
0
accept %
⟨cosθ_P⟩
σ(χ₁₁)
sweeps/s
meas
0

Lattice

Periodic cubic L³. Changing L reinitializes a hot (random) field.
4

Coupling & update

1.00
0.50
0

Run protocol

80
120
5
Thermalization sweeps are discarded; then ⟨cosθ_P⟩ is sampled every `interval` sweeps.

β-scan (confinement headline)

Idle. Sweeps β and plots ⟨cosθ_P⟩ & the Creutz string tension σ(β).

Wilson loop diagram

1
1
4
The rectangle highlighted on the slice is the Wilson loop W(R,T).

Live lattice slice — plaquette action density 1−cos θ_P

low energyhigh energy loop outline
W(R,T)
V(R) = −ln W(R,T+1)/W(R,T)
χ(R,T) string tension
β/2 check (strong cpl.)
Physics notes & correctness checks
Action: S = β·Σ_P [1 − cos θ_P], plaquette phase θ_P = θ_μ(n)+θ_ν(n+μ̂)−θ_μ(n+ν̂)−θ_ν(n) (counterclockwise product).
ΔS per Metropolis move depends only on the 4 staples touching the link (2 directions × forward+backward), never a full-action recompute.
Strong-coupling check: at small β, ⟨cosθ_P⟩ → β/2 (shown as the dashed overlay on the scan plot). At large β, ⟨cosθ_P⟩ → 1−1/(3β) (free-field weak-coupling, d=3).
Confinement: 2+1D compact U(1) confines for all β (Polyakov). The Creutz ratio χ(R,T) = −ln[W(R+1,T+1)W(R,T)/(W(R+1,T)W(R,T+1))] → string tension σ, nonzero and decreasing with β. This is emergent from the geometry/dynamics — nothing is hardcoded.
Auto-tune holds Metropolis acceptance near ~50%; the equilibration plateau appears in the history plot.

(a) plaquette history — equilibration plateau

(b) ⟨cosθ_P⟩ vs β — with β/2 (strong-cpl.) overlay

(c) static potential V(R) vs R

(d) Creutz string tension σ vs β