Driven-Dissipative Spin Glass

• Driven-dissipative spin glasses are nonequilibrium many-body systems characterized by random couplings, external drive, and engineered dissipation, leading to glassy order and replica symmetry breaking.
• They are realized in platforms like cavity QED and cold atom setups using switching fields and Lindblad dynamics for precise experimental control and measurement.
• Controlled drive protocols and tailored dissipation enable the observation of ultrametric state-space structures, enhanced associative memory, and scaling laws in nonthermal regimes.

A driven-dissipative spin glass is a frustrated many-body spin system subject to continuous external driving and engineered dissipation, typically realized in cavity quantum electrodynamics (QED), cold atom platforms, or classical networks with switching fields. Unlike equilibrium spin glasses, which are governed by Boltzmann-Gibbs statistics, driven-dissipative spin glasses operate under nonequilibrium conditions where the interplay of coherent dynamics, external drive protocols, and tailored loss channels gives rise to new forms of glassy order, memory capacity, ultrametricity, and pattern recognition. Experimental realizations have demonstrated the emergence of fundamental phenomena predicted by mean-field spin glass theory—such as replica symmetry breaking (RSB) and ultrametric state-space structure—under conditions far from thermal equilibrium, and have exploited these features for associative memory and novelty detection.

1. Model Architectures and Hamiltonians

Driven-dissipative spin glasses are constructed on Ising ($s_i = \pm 1$) or vector ($\mathbf{S}_i$) degrees of freedom with quenched, randomly signed couplings $J_{ij}$ and are typically subjected to time-dependent fields or photon-mediated cavity interactions. Several paradigmatic models include:

• Spin glass with switching external fields: Systems of $N$ Ising spins with random couplings and a finite set of quenched external fields $\{\mu^{(k)}\}$, periodically switched with interval $\tau$, leading to the instantaneous Hamiltonian

$$E(\mathbf{s};\,\mu^{(k)}) = \sum_{i<j} J_{ij}\,s_i\,s_j + \sum_{i=1}^N \mu_i^{(k)} s_i$$

(Gold et al., 2019).

• Quantum-optical spin glasses: Ensembles of ultracold atoms in optical tweezers, coupled via all-to-all, sign-random Ising or vector interactions $J_{ij}$ mediated by multimode optical cavities. The effective Hamiltonian after adiabatic elimination of fast cavity and excited-state dynamics reads, for the Ising (transverse-field) case,

$$\hat{H} = 2 E_r \sum_{i=1}^n \hat{S}_i^z - \frac{\hbar g_0^2 \Omega^2}{\Delta_A^2 |\Delta_C|} \sum_{i,j=1}^n J_{ij} \hat{S}_i^x \hat{S}_j^x$$

Dissipation is captured with Lindblad collapse operators linear in the collective spin basis (Marsh et al., 28 May 2025).

• $\pm J$ driven Ising glass in external magnetic field: A system on a cubic lattice, with bimodal bond distribution and a uniform field $\mathbf{S}_i$0 swept either sinusoidally or adiabatically. The dissipative dynamics and hysteresis under field driving probe the glassy relaxation modes (Sarıyer et al., 2012).
• Driven-dissipative vector spin glasses: Featuring $\mathbf{S}_i$1-like (rotor) degrees of freedom and vector couplings ($\mathbf{S}_i$2) in confocal cavities. The energy depends on both $\mathbf{S}_i$3 and $\mathbf{S}_i$4, reflecting the vector nature of the glass (Kroeze et al., 2023).

The essential ingredients are both frustration (random, sign-changing couplings) and nonequilibrium drive (external fields, cavity pumps) together with controlled dissipation channels (thermal noise, photon loss, engineered Lindblad terms) that prevent simple thermalization.

2. Dynamics and Dissipation Protocols

Driven-dissipative spin glasses achieve their characteristic glassy behavior through a combination of stochastic or deterministic spin-flip dynamics and continuous or pulsed driving:

• Spin-flip thermalization: Continuous-time Glauber or Arrhenius-like rules for Ising flips, e.g.,

$\mathbf{S}_i$5

with $\mathbf{S}_i$6 (Gold et al., 2019).

• Field-switching protocols: Piecewise-constant or smoothly ramped pump fields, driving the system through ordering transitions with tunable ramp rates ($\mathbf{S}_i$7), affecting the number and accessibility of minima in the glassy landscape (Marsh et al., 28 May 2025).
• Lindblad and quantum trajectory formalism: Cavity-mediated systems employ master equations for the density matrix $\mathbf{S}_i$8, with photon loss giving rise to dissipative collapse operators $\mathbf{S}_i$9. Unraveling these equations yields quantum (or semiclassical) trajectories that sample the nonequilibrium ensemble (Marsh et al., 2023).
• Memory and adaptation: Under continual driving, the system evolves towards configurations (or attractors) that minimize some measure of dissipation—the average work absorbed from the drive, or steady-state entropy production—while retaining statistical memory of the drive history (Gold et al., 2019).

The interplay between timescales—drive switching interval, intrinsic glassy relaxation, measurement times—dictates whether the system exhibits fast adaptation, long-lived memory, or glassy slow dynamics.

3. Emergent Nonequilibrium Phenomena: Memory, RSB, Ultrametricity

Driven-dissipative spin glasses display distinctive nonequilibrium phenomena not found in purely equilibrium systems:

• Self-organized novelty detection: When subjected to a finite set of repeatable, nonuniform external fields, a driven-dissipative spin glass evolves into configurations that absorb minimal work from that particular drive set. The resulting steady states act as novelty detectors, recognizing and responding to novel drive patterns by increased dissipation before adapting again (Gold et al., 2019).
• Replica symmetry breaking (RSB): Despite their nonequilibrium nature, these systems exhibit full RSB in the sense of Parisi, manifested in highly nontrivial distributions of state overlaps ($J_{ij}$0, Parisi function $J_{ij}$1), with interior weight and multiple peaks, measured across repeated identical preparation cycles ("replicas") (Marsh et al., 28 May 2025, Marsh et al., 2023, Kroeze et al., 2023).
• Ultrametric state-space organization: The overlap matrices $J_{ij}$2 among replicas exhibit block-hierarchical ("tree-like") structure. Ultrametricity is demonstrated by statistics such as the $J_{ij}$3-correlator, with distances $J_{ij}$4 obeying the strong triangle inequality. This tree-like organization is observed both in Ising and in vector (XY) driven-dissipative glasses (Marsh et al., 28 May 2025, Kroeze et al., 2023).
• Nonthermal order parameter distributions: The Parisi overlap distributions in driven-dissipative quantum glasses differ qualitatively from their equilibrium analogs, reflecting the nonthermal, measurement-induced stabilization of nonequilibrium glassy order (Marsh et al., 2023).
• Associative memory beyond the Hopfield limit: In quantum-optical driven-dissipative spin glasses, nonequilibrium dynamics (steepest-descent-type updates and polaronic motion) turn spurious minima into reliable attractors. This results in memory capacities surpassing the classical Hopfield bound by factors of up to $J_{ij}$5 for $J_{ij}$6 spin networks (Marsh et al., 15 Sep 2025).

4. Experimental Realizations and Measurement Schemes

The experimental platforms for driven-dissipative spin glasses typically combine ultracold atomic ensembles, optical cavities with engineered multimode structure, and holographic imaging for real-time measurement of spin states:

• Cavity QED platforms: Arrays of $J_{ij}$7 ultracold $J_{ij}$8Rb ensembles (up to $J_{ij}$9), trapped by optical tweezers in near-planar or confocal multimode Fabry–Pérot cavities. Pump ramps are controlled to explore glassy transitions (Marsh et al., 28 May 2025, Kroeze et al., 2023).
• Measurement and imaging: Holographic detection of forward-scattered cavity light allows direct imaging of all $N$0, including real-time phase and amplitude readout at each spin site (Marsh et al., 28 May 2025, Kroeze et al., 2023).
• Replica construction: Multiple identical repetitions of the disorder realization and drive protocol yield independent replicas for overlap statistics; convergence is assessed via bootstrap and Hellinger distances (Kroeze et al., 2023).
• Novelty detection protocols: Alternating among distinct field sets and measuring work absorption rates $N$1 tests the system's memory and detection of environmental changes (Gold et al., 2019).

Performance metrics include the Parisi order parameter $N$2, Edwards-Anderson parameter $N$3, the $N$4-correlator for ultrametricity, memory overlap distributions, and the Shannon entropy of the observed spin configuration distribution.

5. Scaling Laws, Dynamic Regimes, and Physical Interpretation

Driven-dissipative spin glasses exhibit scaling laws and dynamic regimes characteristic of both glass physics and nonequilibrium statistical mechanics:

• Adiabatic vs lagging regimes: In the $N$5 Ising glass under periodic magnetic field drive, the hysteresis area $N$6 serves as a measure of dissipative loss. For frequencies $N$7 (set by intrinsic relaxation), $N$8, corresponding to the adiabatic limit. For $N$9, $\{\mu^{(k)}\}$0 grows as a power law in $\{\mu^{(k)}\}$1, with exponent $\{\mu^{(k)}\}$2 increasing with glassiness ($\{\mu^{(k)}\}$3 deep in the glass) (Sarıyer et al., 2012).
• Entropy control and accessible minima: The number and diversity of occupied minima (measured by entropy $\{\mu^{(k)}\}$4) can be tuned by the ramp rate through the ordering transition; slow ramps yield few deep minima, fast ramps populate a large ensemble of shallow states (Marsh et al., 28 May 2025).
• Role of quantum entanglement: In quantum trajectories, transient many-body entanglement allows access to lower-energy glassy states unachievable by semiclassical paths, directly linking quantum measurement backaction to RSB (Marsh et al., 2023).
• Physical interpretation: The characteristic observables (overlap distribution, ultrametric $\{\mu^{(k)}\}$5-correlator, work-absorption memory) provide direct, experimental access to underlying archetypes of glassy organization.

A summary of observable regimes:

Drive Protocol	Observables	Phenomena
Field-switching, slow	$\{\mu^{(k)}\}$6, $\{\mu^{(k)}\}$7, $\{\mu^{(k)}\}$8	Novelty detection, memory
Pump ramp, variable	$\{\mu^{(k)}\}$9, $\tau$0, $\tau$1	Entropy tuning, RSB
Sinusoidal driving	$\tau$2, scaling expts.	Dynamic hysteresis, scaling

6. Theoretical and Practical Implications

Driven-dissipative spin glasses provide a platform for investigating complex systems well beyond standard thermal equilibrium, enabling:

• Testing universality of glassy orders: Experiments demonstrate that full RSB and ultrametric hierarchy manifest under nonequilibrium, nonthermal, and quantum conditions, supporting the robustness of these orders beyond mean-field equilibrium theory (Marsh et al., 28 May 2025, Kroeze et al., 2023).
• Associative memory and neuromorphic computing: Spin glass networks with programmable $\tau$3, driven and read out optically, can function as high-capacity content-addressable (Hopfield-type) memories, where even spurious states are turned into reliable retrieval basins (Marsh et al., 15 Sep 2025).
• Aging, rejuvenation, and learning: Fine control over drive and measurement enables exploration of glassy aging, rejuvenation, and precursor forms of learning due to atomic or network plasticity (Marsh et al., 28 May 2025, Marsh et al., 15 Sep 2025).
• Quantum simulation of nontrivial optimization: The realization and characterization of nonequilibrium RSB opens routes to simulating NP-hard problems and investigating glassy optimization in quantum hardware.

A plausible implication is that driven-dissipative spin glasses, both classical and quantum, may become generic testbeds for studying the interface of stochastic thermodynamics, quantum measurement theory, and complex adaptive behavior.

7. Future Directions and Open Problems

Prospective research avenues in driven-dissipative spin glasses include:

• Scaling up system size and connectivity: Expanding to larger arrays ($\tau$4), engineering interaction graph topologies and disorder profiles, and probing finite-size crossover in RSB and ultrametricity (Marsh et al., 28 May 2025, Kroeze et al., 2023).
• Quantum regime and spin-$\tau$5 systems: Moving from ensemble "large-$\tau$6" pseudospins to genuine spin-$\tau$7 via Rydberg-blockaded arrays, enabling direct access to quantum SK models (Marsh et al., 28 May 2025).
• Altered dissipation engineering: Controlling Lindblad channels to explore the impact of measurement backaction and decoherence structure on glassy orders (Marsh et al., 2023).
• Optimization and learning dynamics: Systematically studying "polaronic" effects and feedback between network dynamics and connectivity as analogs of short-term synaptic plasticity (Marsh et al., 15 Sep 2025).
• Comparison between equilibrium and nonequilibrium glassiness: Quantifying similarities and departures in overlap landscapes, metastable state statistics, and dynamical aging across thermal and driven settings (Kroeze et al., 2023).

The realization of driven-dissipative spin glasses in cavity QED and related systems establishes a new regime of complex matter where the classical paradigms of glass and disorder are enriched by quantum trajectories, measurement, and engineered environmental couplings.