Abstract
We consider two schematic models of glasses subjected to oscillatory shear deformation, motivated by the observations, in computer simulations of a model glass, of a nonequilibrium transition from a localized to a diffusive regime as the shear amplitude is increased, and of persistent memory effects in the localized regime. The first of these schematic models is the NK model, a spin model with disordered multi-spin interactions previously studied as a model for sheared amorphous solids. The second model, a transition matrix model, is an abstract formulation of the manner in which occupancy of local energy minima evolves under oscillatory deformation cycles. In both of these models, we find a behavior similar to that of an atomic model glass studied earlier. We discuss possible further extensions of the approaches outlined.
Introduction
Amorphous materials—ranging from metallic glasses to colloids and foams—exhibit a fascinating dual nature when subjected to cyclic stress. At low strain amplitudes, these systems eventually settle into a "localized" regime, where they return to the same configuration after each cycle. However, once a critical strain threshold ($\gamma_c$) is crossed, the material enters a "diffusive" regime, wandering through its energy landscape without returning to a fixed state.
Beyond this phase transition, these materials exhibit a form of "memory." If a glass is "trained" by repeated cycles at a specific amplitude, it retains a signature of that amplitude. This memory can be retrieved by probing the system at varying amplitudes and observing specific discontinuities, or "kinks," in physical observables like particle displacement. While these phenomena have been documented in complex atomic simulations, the underlying mechanisms remain obscured by the complexity of many-body interactions. This study utilizes two simplified "schematic" models to identify the minimal ingredients necessary for these memory effects.
Methods: Abstracting the Energy Landscape
To isolate the fundamental drivers of memory and phase transitions, the researchers employed two distinct modeling approaches:
The NK Model
The NK model, originally developed for evolutionary biology, is adapted here as a spin model representing a glass. It consists of $N$ sites, each in a binary state (0 or 1). The energy of each site depends on its own state and the states of $K$ neighbors, creating a tunable "rugged" energy landscape. By introducing a strain parameter ($\gamma$) into the energy function, the authors simulate Athermal Quasi-Static (AQS) deformation. In this process, the system is nudged by a change in strain and then allowed to relax to the nearest local energy minimum via steepest descent.
The Transition Matrix (TM) Model
The TM model abstracts the process further by ignoring the physical coordinates of the energy landscape. Instead, it treats the deformation cycle as a mathematical mapping between discrete energy minima (inherent structures). If the system starts in state $A$, a full cycle of strain $\gamma$ maps it to state $B$. This is represented by a sparse transition matrix $P$. This approach allows for a rigorous classification of states into "absorbing" (stable) or "recurring" (part of a cycle).
Key Results
The study demonstrates that these simplified models successfully replicate the complex behaviors of atomic glasses.
The Localized-to-Diffusive Transition
Both models reproduce a sharp transition from reversible to irreversible behavior at a critical amplitude ($\gamma_c \approx 0.5$). Below this threshold, the system eventually finds an "absorbing" state—a configuration that is stable under the given cyclic strain. Above the threshold, the system remains in a diffusive state, continuously exploring new configurations.
Memory Signatures and System Size
The "memory" of the training amplitude is encoded in the specific energy minimum the system occupies. Training at an amplitude $\gamma_{train}$ forces the system into a state that is stable specifically for that amplitude. When the system is later probed, the Mean Squared Displacement (or Hamming distance $d$ in the spin model) shows a sharp "kink" or discontinuity at the training amplitude.
As shown in the results, the transition becomes increasingly well-defined as the system size ($N$ or the number of states $M$) increases, suggesting a true phase transition in the thermodynamic limit. Furthermore, the models demonstrate the ability to store multiple memories simultaneously by alternating training amplitudes, resulting in multiple distinct kinks during the readout phase.
Discussion and Future Directions
The success of the NK and TM models suggests that memory in glasses is a universal property of driven disordered systems navigating rugged landscapes, rather than a specific consequence of atomic-scale physics. However, the models currently lack spatial correlations; in real glasses, atoms move in coordinated clusters. Future work aims to incorporate a notion of "distance" or structural organization between states in the Transition Matrix model to better capture the spatial nuances of physical amorphous solids.
Reference: Fiocco, D., Foffi, G., & Sastry, S. (2021). Memory effects in schematic models of glasses subjected to oscillatory deformation. arXiv:1503.00478v1.
Summary updated