Finite Edwards-Anderson Order Parameter

• Finite Edwards-Anderson order parameter is defined through spatial and disorder averaging of squared local observables, capturing persistent frozen order in disordered phases.
• It distinguishes glassy states from homogeneous phases by measuring overlaps and local density fluctuations via experimental and simulation techniques.
• Finite-size corrections and scaling analyses are crucial, using methods like Monte Carlo sampling and field-theoretic approaches to assess the robustness of glassy order.

The finite Edwards-Anderson (EA) order parameter quantifies the extent of frozen, random order in disordered systems, particularly spin glasses and related models. It serves as a central diagnostic to distinguish true glassy phases from homogeneous or thermally fluctuating phases, both in classical and quantum contexts. Finite values of the EA order parameter reflect persistent local order that survives both disorder and ensemble averaging, marking regimes of replica symmetry breaking or glass rigidity. Techniques and definitions vary by physical system, but all constructions involve spatial averaging and disorder ensembles, directly linking q_{EA} to physical observables such as overlaps, local densities, or spectral linewidths.

1. Formal Definitions of the EA Order Parameter

The EA order parameter is generally defined in terms of the square of local observables, averaged over both thermal fluctuations and disorder. For an Ising spin glass of N spins with variables $S_i = \pm1$, the conventional finite-size EA parameter is

$$q_{\rm EA}(L,T) = \frac{1}{N}\sum_{i=1}^{N}\bigl[\langle S_i\rangle^2\bigr]_{\rm av}$$

but, operationally, it is often extracted as the position of the dominant peak of the overlap distribution $P(q)$, where the overlap between two replicas $\alpha$ and $\beta$ of the same disorder realization is

$q = \frac{1}{N}\sum_{i=1}^N S_i^\alpha S_i^\beta$

and

$q_{\rm EA}(L,T) = \arg\max_{q>0} P(q)$

In quantum or bosonic contexts, such as the disordered Bose-Hubbard model, the EA order parameter for the Bose glass is

$$q_{\rm EA} = \overline{\,\langle \hat n_i\rangle^2\,} - \overline{\langle \hat n_i\rangle}^2 \equiv \overline{\bigl(\langle \hat n_i\rangle - \overline{\langle \hat n_i\rangle}\bigr)^2}$$

where the overbar denotes a disorder average, and angular brackets denote a thermal or mean-field average (Thomson et al., 2016). In random-field Ising-nematic models relevant to FeSe, the EA parameter is

$$q_{\rm EA} = \frac{1}{N} \sum_{i} [ \langle \tau_i^z \rangle^2 ]_{\text{dis}} - \bar\phi^2$$

with $\tau_i^z$ a local Ising-nematic pseudospin, and $\bar\phi$ the site-averaged nematic moment (Wiecki et al., 2021).

2. Physical Significance Across Systems

A finite $q_{\rm EA}$ signifies rigid local order that is locked to the randomness of the disorder rather than to a uniform global symmetry-broken state. In spin glasses, nonzero $q_{\rm EA}$ below the critical temperature $T_c$ demarcates the spin-glass phase and reflects frozen-in randomness in the spin configuration (Lewenstein et al., 2022). In the Bose glass, $q_{\rm EA} > 0$ arises from localized superfluid puddles pinned by disorder, contrasting the uniform Mott insulator and clean superfluid, where $q_{\rm EA} = 0$ (Thomson et al., 2016). In random-field Ising-nematic systems, a finite $q_{\rm EA}$ directly correlates with the NMR spectral width due to local nematic domains (Wiecki et al., 2021).

Characteristic features:

• $q_{\rm EA} = 0$ in clean, uniform phases or above $T_c$.
• $q_{\rm EA} > 0$ only when disorder induces persistent inhomogeneities or glassy clusters.
• At criticality or near the transition, $q_{\rm EA}$ vanishes as a critical exponent.

3. Measurement Methodologies for Finite-Size q_{EA}

Spin Glasses

Monte Carlo simulation protocols for the 3D Edwards-Anderson model utilize parallel tempering with several thousand disorder samples. The overlap $q$ is computed from thermally independent replicas, the $P(q)$ distribution is histogrammed, and $q_{\rm EA}(L,T)$ is defined as the location of the main peak at $q>0$ (Yucesoy et al., 2012). The value of $q_{\rm EA}(L)$ decreases with system size in the EA model, with reported values at $T=0.42$ for $L=4$ to $L=12$ ranging from $\sim$0.65 to $\sim$0.45.

Quantum/Bosonic Systems

In the Bose glass, quantum gas microscopes can detect the EA parameter through single-shot parity-resolved snapshots. By reconstructing local densities $\langle n_i \rangle$ over many disorder patterns, one computes

$$q_{\rm EA} = \frac{1}{N_{\mathrm{sites}}}\sum_i\left[ \langle n_i \rangle^2_{\rm dis} - (\langle n_i \rangle_{\rm dis})^2 \right]$$

Approximate $q_{\rm EA}$ estimates are possible using circular averaging for a single disorder realization, but with systematic biases (Thomson et al., 2016).

Field-Theoretic and Analytical Approaches

Saddle-point analysis in the Haake-Lewenstein-Wilkens framework expresses $q_{\rm EA}$ in terms of the squared local magnetization in an effective mean-field theory, yielding $q_{\rm EA} = m^2$ with $m$ from a bulk self-consistency equation. In three and four dimensions, $q_{\rm EA} > 0$ for $T < T_c$, but in two dimensions the method overestimates $T_c$ due to strong boundary effects (Lewenstein et al., 2022).

Lee–Yang Zeros

The density and scaling of overlap-field Lee–Yang zeros provide an independent estimator: the initial slope of the integrated zero density $G(\epsilon)$ in the spin-glass phase yields $q_{\rm EA} = (\pi/\beta)a_1$, and this agrees within 5% with overlap-based estimates up to $L=32$ (Baños et al., 2012).

Eigenstate Spin-Glass Order Parameter (ESG)

For many-body localized spin-glass order, reduced density matrices of two sites characterize $q_{\rm EA}$ analogues in both static and dynamic regimes. The conventional $q_{\rm EA}$ is compared with the ESG parameter derived from two-site reduced density matrices; both track each other in disorder-driven transitions and in finite-size scaling (Javanmard et al., 2018).

4. Finite-Size Corrections and Scaling

Finite-size $q_{\rm EA}(L, T)$ exhibits downward drift with increasing $L$; no explicit scaling exponent for the correction is universally reported (Yucesoy et al., 2012). In glassy random energy models or 1/f noise models, the exact representation yields

$$q_{\rm EA}(\beta, N) = q_{\rm EA}(\beta, \infty) + \frac{c_1(\beta)}{\ln N} + \frac{c_2(\beta)}{(\ln N)^2} + \ldots$$

with $q_{\rm EA}(\beta, \infty)$ analytic in the ergodic phase and singular at the glass transition (Cao et al., 2016). In EA spin glasses, the broadening of the main $P(q)$ peak narrows and sharpens with $N$, but the position $q_{\rm EA}(L)$ scales slowly; in the SK model, many additional peaks at lower $|q|$ emerge characteristic of full RSB, but in the 3D EA model, $\Delta(q_0, \kappa)$ for peaks at small $q$ remains small and flat, supporting a two-state scenario.

5. System-Dependent Behavior and Experimental Probes

System	Definition of $q_{\rm EA}$	Experimental Proxy/Observable
3D Ising EA Spin Glass	$P(q)$ peak position	Overlap distribution from MC simulations
Disordered Bose-Hubbard	Density variance across sites	QGM snapshots, parity-resolved imaging
Random-Field Ising-Nematic	$[\langle τ_i^z \rangle^2]_{dis}$	NMR spectral width, local moments
Circular 1/f-Noise Model	Replica/Jack polynomial sum	Moments of Gibbs weights, finite-$N$ scaling

NMR experiments measure $\Delta\nu(T,P) = \alpha \sqrt{q_{\rm EA}(T)}$, directly probing $q_{\rm EA}$ above the nematic transition (Wiecki et al., 2021). In ultracold atom experiments, the BG phase can be unambiguously detected by the rapid onset of $q_{\rm EA} \approx 0.2-0.25$ with disorder, sharply discriminating it from MI and SF states (Thomson et al., 2016).

6. Controversies and Limitations

The Haake-Lewenstein-Wilkens approach, while deriving $q_{\rm EA}(T)$ analytically for any dimension, overestimates the existence of a finite-$T$ spin-glass phase in $d=2$ due to neglect of dominant boundary or domain-wall effects, highlighting the limitations of mean-field approximations in low dimensions (Lewenstein et al., 2022). In 3D, both numerical and analytical work support finite $q_{\rm EA}$ below $T_c$ and a two-state (droplet/chaotic-pairs) scenario, in sharp contrast to the infinite-state structure of the SK model (Yucesoy et al., 2012). The finite-size behavior, slow drift with $L$, and system-specific corrections remain the subject of ongoing investigation. In quantum glass settings, the distinction between thermal and disorder-induced fluctuations in finite $q_{\rm EA}$ is subtle and demands high-precision, disorder-ensemble-resolved measurement protocols (Thomson et al., 2016, Javanmard et al., 2018).

7. Broader Implications and Extensions

A finite EA order parameter functions as a universal marker of quenched glassy order across a diverse range of systems, including spin glasses, bosonic glasses, random-field nematics, and log-correlated disordered landscapes (log-REM, 1/f-noise). In each context, $q_{\rm EA}$ encodes the persistent memory of frozen, disorder-pinned inhomogeneities and underpins the experimental identification of glass phases. In the quantum domain, $q_{\rm EA}$ and its ESG generalizations provide scalable diagnostics that extend into non-equilibrium and many-body localized regimes, with potential application in quantum simulation platforms (Javanmard et al., 2018). The rigorous determination of $q_{\rm EA}$ and its systematics remains central to the understanding of glassy matter, replica symmetry breaking, and the interplay between disorder, interactions, and topology in complex materials.