Emergent Loop Geometry in Gauge-Equivariant Graph Neural Networks

Ali Rayat; Gia-Wei Chern

Home/Archive/arXiv

arXiv · 2610.18742

Emergent Loop Geometry in Gauge-Equivariant Graph Neural Networks

Ali Rayat·Gia-Wei Chern

View original source ↗Read source document ↗Export metadata ↓

Abstract

Gauge-equivariant graph neural networks generate ordered products of gauge links through local message passing, but the geometry of these internal structures is not captured by the receptive field alone. We investigate the statistical ensemble of closed lattice walks generated by recursive node- and edge-feature operations, independently of network training. Analytical and numerical results reveal a separation between rapid growth of contour length and much slower spatial expansion: deeper compositions generate increasingly long paths through repeated traversal of relatively compact regions. Loop expansions of fermionic observables provide a physical interpretation of this hierarchy, showing how spatial locality remains compatible with high-order dependence on nearby gauge fields. These findings identify complementary roles of network depth in extending spatial reach and enriching the representations constructed within that reach, providing a geometrical perspective on learning lattice gauge observables.

Explore related subjects

cond-mat.stat-mech cond-mat.str-el cs.LG hep-lat

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Ali Rayat, Gia-Wei Chern. 2026-10-13. Emergent Loop Geometry in Gauge-Equivariant Graph Neural Networks. https://arxiv.org/abs/2610.18742

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Earthquake Dynamics as a Benchmark for Scale-Invariant Behavior

We analyze regional earthquake energy statistics from the Southern California and Japan seismic catalogs and find scale-invariant energy distributions characterized by an exponent $τ\simeq 1.67$. To quantify how closely scale-invariant dynamics with different exponent values resemble real earthquakes, we generate synthetic energy distributions over a wide range of $τ$ under conditions of constant activity. Earthquake-like behavior, in a broad sense, is obtained for $1.5 \leqslant τ< 2.0$. When energy variations are further restricted to be within a factor of ten relative to real earthquakes, the admissible range narrows to $1.58 \leqslant τ\leqslant 1.76$. We identify the physical mechanisms governing the dynamics in the different regimes: fault dynamics characterized by a balance between slow energy accumulation and release through scale-free events in the earthquake-like regime; externally supplied energy relative to a slowly driven fault for $τ< 1.5$; and dominance of small events in the energy budget for $τ> 2$

cond-mat.stat-mech↗

arXiv2026

Higher-order response theory in optimal stochastic thermodynamics

Linear response theory has found many applications in statistical physics. One of these is to compute minimal-work protocols that drive nonequilibrium systems between different thermodynamic states, which are useful for designing engineered nanoscale systems and understanding biomolecular machines. We clarify the relationships between linear-response-based approximations used to study optimal protocols in different driving regimes by demonstrating how they arise from a truncation of a causal response, or Volterra, series expansion. We then construct higher-order response terms and discuss the utility and drawbacks of their inclusion in the determination of work-minimizing protocols. We focus on the cases of an overdamped particle in a harmonic trap, and a double well system modeled as a two-state discrete system, ultimately showing that while the inclusion of higher-order response in work approximations can improve predictions of excess work and modestly improve the resulting higher-order optimal protocols, a significant computational expense is incurred while introducing the possibility of predicting arbitrarily low and unphysical negative excess work.

cond-mat.stat-mech↗

arXiv2026

Active fluctuations drive inertia-free absolute negative mobility in soft and living matter

Absolute negative mobility (ANM) is one of the most paradoxical transport phenomena in which a setup moves on average in a direction opposite to the applied force. The simplest system exhibiting this effect in minimal one-dimensional dynamics involves an inertial particle subjected to a constant bias while dwelling in a nonlinear, symmetric periodic potential in a nonequilibrium, nonstationary state generated by external driving. The necessity of inertia poses a significant conceptual barrier in predicting the ANM in a microscopic world of soft and living matter such as biological cells. In this work, we solve a two-decade old puzzle and demonstrate that none of the previously thought factors required for this effect are decisive. We show that the ANM may occur in a system composed of an overdamped particle in a piecewise linear symmetric periodic potential in an equilibrium state, provided that it is additionally driven by active fluctuations in the form of white Poisson shot noise. Our result may help to explain exotic transport behaviour emerging in biological cells where dynamics is typically overdamped and assisted by active fluctuations derived from various metabolic activities. It can also be exploited for effective separation strategies in a microscopic world, thus transforming fluctuations from a nuisance into a functional resource.

cond-mat.stat-mech↗