Mesoscopic real space structures in aging spin-glasses: the Edwards-Anderson model

TL;DR

This paper investigates the aging dynamics of the 3D Edwards-Anderson spin glass model, revealing that non-equilibrium events called quakes form a Poisson process, with cluster growth and temperature scaling linked to experimental phenomena.

Contribution

It introduces a novel interpretation of spin glass aging based on record dynamics and Poisson quakes, connecting real space structures to thermodynamic and experimental observations.

Findings

Quakes are statistically independent and follow a Poisson process.

Cluster sizes grow logarithmically over time.

Temperature scaling with T^{1.75} collapses data across different conditions.

Abstract

Isothermal simulational data for the 3D Edwards-Anderson spin glass are collected at several temperatures below $T_{\rm c}$ and, in analogy with a recent model of dense colloidal suspensions,interpreted in terms of clusters of contiguous spins overturned by quakes, non-equilibrium events linked to record sized energy fluctuations. We show numerically that, to a good approximation, these quakes are statistically independent and constitute a Poisson process whose average grows logarithmically in time. The overturned clusters are local projections on one of the two ground states of the model, and grow likewise logarithmically in time. Data collected at different temperatures $T$ can be collapsed by scaling them with $T^{1.75}$, a hitherto unnoticed feature of the E-A model, which we relate on the one hand to the geometry of configuration space and on the other to experimental memory and rejuvenation effects. The rate at which a cluster flips is shown to decrease exponentially with the size of the cluster, as recently assumed in a coarse grained model of dense colloidal dynamics. The evolving structure of clusters in real space is finally sssociated to the decay of the thermo-remanent magnetization. Our analysis provides an unconventional coarse-grained description of spin glass aging as statistically subordinated to a Poisson quaking process and highlights record dynamics as a viable common theoretical framework for aging in different systems.