Echoes in the Void: Unmasking Hidden Dynamics in Many-Body Localization

Author: Denis Avetisyan


New research reveals that localized resonances profoundly affect the long-term behavior of disordered quantum systems, challenging conventional interpretations of imbalance measurements.

Despite an overall appearance of saturation within the many-body localized regime, detailed analysis of site-resolved spin autocorrelators-calculated via unitary time evolution governed by <span class="katex-eq" data-katex-display="false">\text{Eq. (5)}</span> and averaged over time as defined in <span class="katex-eq" data-katex-display="false">\text{Eq. (7)}</span>-reveals localized resonant dynamics at individual sites, highlighted against the backdrop of a seemingly uniform imbalance <span class="katex-eq" data-katex-display="false">{\cal{I}}^{m}(t)</span> as described by <span class="katex-eq" data-katex-display="false">\text{Eq. (3)}</span>, demonstrating that spatial averaging can obscure critical information about the system’s underlying behavior even in strongly disordered systems with <span class="katex-eq" data-katex-display="false">\Delta = 1</span> and <span class="katex-eq" data-katex-display="false">h = 10</span>.
Despite an overall appearance of saturation within the many-body localized regime, detailed analysis of site-resolved spin autocorrelators-calculated via unitary time evolution governed by \text{Eq. (5)} and averaged over time as defined in \text{Eq. (7)}-reveals localized resonant dynamics at individual sites, highlighted against the backdrop of a seemingly uniform imbalance {\cal{I}}^{m}(t) as described by \text{Eq. (3)}, demonstrating that spatial averaging can obscure critical information about the system’s underlying behavior even in strongly disordered systems with \Delta = 1 and h = 10.

Site-resolved dynamics demonstrate that local resonances impact finite-size scaling and may lead to misinterpretations of experimental data in many-body localized systems.

While many-body localization (MBL) is often characterized by global observables like imbalance, this work-‘Beyond the imbalance: site-resolved dynamics probing resonances in many-body localization’-demonstrates that spatially averaged measurements can obscure crucial microscopic dynamics. Through advanced numerical simulations of the disordered XXZ chain, we reveal resonant structures and rare local instabilities within the MBL phase, manifested in site-resolved spin autocorrelators. These localized effects not only explain finite-size scaling anomalies in long-time imbalance measurements, but also suggest a richer, more nuanced picture of ergodicity breaking than previously understood. Could a refined, site-resolved approach be key to fully unlocking the complexities of MBL and its connection to rare-region phenomena?


The Unraveling of Equilibrium: A Challenge to Conventional Physics

For decades, the prevailing understanding of physics dictated that closed quantum systems, entirely isolated from external influences, would inevitably reach thermal equilibrium – a state of maximum entropy where energy is evenly distributed. However, a growing body of evidence demonstrates this isn’t universally true. Certain quantum systems, specifically those characterized by significant disorder – randomness in their constituent components – exhibit a strikingly different behavior. These systems resist thermalization, retaining a ‘memory’ of their initial conditions and preventing energy from spreading throughout. This unexpected defiance of established principles suggests the existence of a new, fundamentally different phase of matter where interactions and disorder combine to create a localized, non-equilibrium state, challenging long-held assumptions about the fate of isolated quantum systems.

Many-Body Localization (MBL) represents a surprising departure from established principles of quantum mechanics, suggesting the existence of a novel phase of matter distinct from the traditionally understood phases like solids, liquids, or gases. Unlike systems that inevitably succumb to thermalization – a process where energy distributes evenly – MBL systems retain a ‘memory’ of their initial conditions, preventing energy flow and maintaining localized excitations. This isn’t merely a static frozen state; rather, it’s a dynamical yet non-ergodic phase, where interactions and disorder conspire to create a complex landscape that traps quantum information. The implications of MBL extend beyond condensed matter physics, potentially offering pathways to robust quantum information storage and processing, as the localized states are naturally protected from decoherence – a major obstacle in building quantum computers. The very existence of MBL challenges conventional wisdom about how complex quantum systems behave, pushing the boundaries of theoretical understanding and opening new avenues for exploration in the quantum realm.

Investigating Many-Body Localization (MBL) necessitates discerning how a disordered quantum system retains a “memory” of its initial conditions, a surprisingly complex undertaking. Unlike typical thermalizing systems where information rapidly disperses, MBL systems exhibit an unusual persistence of this initial state information – but accessing this ‘memory’ is hindered by the system’s inherent strong interactions and randomness. Recent studies suggest this preservation isn’t uniform; instead, it appears linked to the formation of local resonances within the disordered potential. These resonances effectively trap and protect certain quantum states, contributing to the observed imbalance between localized and delocalized behavior. Determining the precise role of these resonances – and how they collectively dictate the system’s long-term evolution – remains a central challenge in unraveling the mysteries of MBL and its implications for a novel phase of matter.

Analysis of full exact diagonalization simulations at strong disorder reveals a transition in the dominant many-body resonances from strong polarization (<span class="katex-eq" data-katex-display="false">\mathcal{A} > 0.8</span>) to two-body (<span class="katex-eq" data-katex-display="false">\mathcal{A} \\in [0.53, 0.8]</span>) and finally to primarily three-body resonances (<span class="katex-eq" data-katex-display="false">\mathcal{A} < 0.53</span>), as demonstrated by the redistribution of weight among these regimes.
Analysis of full exact diagonalization simulations at strong disorder reveals a transition in the dominant many-body resonances from strong polarization (\mathcal{A} > 0.8) to two-body (\mathcal{A} \\in [0.53, 0.8]) and finally to primarily three-body resonances (\mathcal{A} < 0.53), as demonstrated by the redistribution of weight among these regimes.

Quantum Imbalance: Decoding the Echoes of Initial State

The quantum imbalance is a measurable quantity used to determine the degree to which a quantum system retains information about its initial state over time. Specifically, it quantifies the overlap between the time-evolved state and the initial state, expressed mathematically as | \langle \psi_0 | \psi(t) \rangle |^2, where \psi_0 represents the initial state and \psi(t) the state at time t. A larger imbalance value indicates stronger retention of initial state information, signifying a reduced degree of thermalization or equilibration. Conversely, a small imbalance suggests the system has largely lost memory of its initial conditions, approaching a statistically mixed state. This observable is particularly useful in studying many-body localization, where the imbalance remains finite even at long times, indicating a failure of thermalization.

Calculating the quantum imbalance necessitates a reduction in the system’s Hilbert space dimensionality. This is commonly accomplished through the application of the partial trace, \text{Tr}_{B}[|\psi\rangle\langle\psi|], which effectively integrates out degrees of freedom associated with subsystem B, leaving only the relevant subsystem A. Alternatively, the full trace, \text{Tr}[|\psi\rangle\langle\psi|], can be employed to obtain a single value representing the total probability retained within the initial state. Both techniques simplify the density matrix and allow for computationally tractable calculations of the imbalance, which quantifies the degree to which the system remembers its initial configuration.

Analysis of the quantum imbalance, used to probe localization, necessitates both time and site averaging to yield interpretable data. Time averaging reduces the impact of dephasing and transient fluctuations, while site averaging mitigates the effects of disorder realization. Research indicates that the finite-size scaling of the averaged imbalance is not universally positive; its sign-positive or negative-is contingent on the specific initial state prepared in the system. This dependence on the initial state suggests that the nature of localization, as revealed by the imbalance, is sensitive to the method of state preparation and subsequent dynamics.

Analysis of the average imbalance <span class="katex-eq" data-katex-display="false">\mathcal{I}_{avg} \approx 0.83</span> reveals that fluctuating sites are rapidly suppressed, the distribution of <span class="katex-eq" data-katex-display="false">\mathcal{I}^m</span> exhibits self-averaging, and secondary peaks in the distribution of <span class="katex-eq" data-katex-display="false">\mathcal{A}_j</span> indicate the presence of local resonances.
Analysis of the average imbalance \mathcal{I}_{avg} \approx 0.83 reveals that fluctuating sites are rapidly suppressed, the distribution of \mathcal{I}^m exhibits self-averaging, and secondary peaks in the distribution of \mathcal{A}_j indicate the presence of local resonances.

The XXZ Chain: A Controlled Descent into Localization

The anisotropic XXZ spin chain, described by the Hamiltonian H = \sum_{i} (S_i^x S_{i+1}^x + S_i^y S_{i+1}^y + \Delta S_i^z S_{i+1}^z), serves as a valuable theoretical platform for investigating Many-Body Localization (MBL). Its tractability stems from being a relatively simple model compared to fully interacting systems, allowing for numerical simulations and analytical approximations. The anisotropy parameter, Δ, controls the relative strength of interactions along the z-axis versus the x-y plane; varying Δ allows researchers to explore the transition between different phases and the emergence of localized behavior. This model’s richness lies in its ability to exhibit a complex interplay between disorder and interactions, crucial for understanding the mechanisms driving MBL, and providing a benchmark for more complex systems.

Introducing random disorder to the XXZ spin chain induces a transition towards a Many-Body Localized (MBL) phase. This localization arises from the disruption of resonant interactions that typically facilitate energy and information transport within the system. Specifically, the introduction of randomness creates a fluctuating potential landscape that inhibits delocalization; the system’s eigenstates cease to span the entire Hilbert space, becoming fragmented and localized around specific configurations. This results in a breakdown of thermalization, where the system fails to reach equilibrium and retains memory of its initial conditions. The strength of the disorder relative to the system’s intrinsic energy scales-specifically, the exchange interaction J-determines the extent of localization and the characteristics of the MBL phase.

Imbalance measurements in the disordered XXZ chain are significantly affected by short-range resonance effects arising from interactions between neighboring spins. These resonances are detectable through analysis of the field difference between adjacent sites; a threshold of 0.08302 in this field difference reliably indicates the presence of a resonant interaction. This threshold value is empirically determined and represents the minimum field disparity required to observe a measurable impact on the system’s localization properties, effectively serving as a signature for local, interacting regions within the otherwise disordered chain.

Full diagonalization of the XXZ Hamiltonian [Eq. (4)] at <span class="katex-eq" data-katex-display="false">h=10</span> and <span class="katex-eq" data-katex-display="false">\Delta=1</span> reveals a complex distribution of the infinite-time, full-trace observable <span class="katex-eq" data-katex-display="false">{\\cal{A}}\\_{j}^{\\rm Full}(\\in fty)</span> [Eq. (9)] differing from individual eigenstate magnetizations, and exhibits resonance peaks at approximately <span class="katex-eq" data-katex-display="false">{\\cal{A}}\\approx 0.603</span> and <span class="katex-eq" data-katex-display="false">{\\cal{A}}\\approx 0.46</span> consistent with theoretical predictions [Eqs. (22) and (23)].
Full diagonalization of the XXZ Hamiltonian [Eq. (4)] at h=10 and \Delta=1 reveals a complex distribution of the infinite-time, full-trace observable {\\cal{A}}\\_{j}^{\\rm Full}(\\in fty) [Eq. (9)] differing from individual eigenstate magnetizations, and exhibits resonance peaks at approximately {\\cal{A}}\\approx 0.603 and {\\cal{A}}\\approx 0.46 consistent with theoretical predictions [Eqs. (22) and (23)].

Initial Conditions and the Ghosts in the Machine: Local Integrals of Motion

The evolution of a many-body localized (MBL) system is acutely sensitive to its initial configuration, meaning even subtle differences in the starting state can lead to dramatically different long-term behavior and measurable imbalances. Consider a system initialized in a Néel state – an antiferromagnetic arrangement where neighboring spins alternate – versus one containing domain walls, boundaries between regions of differing magnetic orientation. The Néel state, being highly ordered, exhibits a distinct dynamical response compared to a state riddled with defects like domain walls, which introduce localized disruptions and alter the pathways of information propagation. This sensitivity isn’t merely a theoretical curiosity; it reflects the fundamental nature of MBL, where localization constrains dynamics and amplifies the impact of initial conditions, effectively ‘remembering’ the starting point throughout its evolution and influencing observable quantities.

Within the Many-Body Localization (MBL) phase, a system’s dynamics are fundamentally restricted by the existence of local integrals of motion. These are conserved quantities that are, crucially, localized to small regions of the system, preventing information from spreading and effectively halting thermalization. Unlike in typical interacting systems where energy can flow freely, the presence of these localized constraints means that the system evolves within a fragmented space of states, dictated by these immobile, conserved quantities. Consequently, the system does not explore all possible states, leading to a suppression of diffusion and a breakdown of ergodicity-the system remains ‘stuck’ in a limited portion of its Hilbert space. The strength of this localization and the specific form of these integrals dictate the precise nature of the non-ergodic behavior and are central to understanding the unique properties of the MBL phase, including the absence of heat flow and the long-lived quantum memories it enables.

The intricate behavior of many-body localized (MBL) systems can be illuminated through the application of quench dynamics – a sudden alteration of the system’s governing parameters. This technique doesn’t merely disrupt the system; it acts as a sensitive probe of the underlying structure of local integrals of motion, those conserved quantities which dictate the allowed dynamics within the MBL phase. Analysis of the time-averaged autocorrelator, a measure of how the system ‘remembers’ its initial state, reveals a characteristic relationship – Z ∝ 𝒢²/ (1 + 𝒢²) – where 𝒢 represents a parameter quantifying the strength of interactions between these local integrals. This functional form provides a direct window into the system’s hidden order, allowing researchers to map the landscape of these conserved quantities and understand how they constrain the system’s evolution following the quench.

The long-time imbalance exhibits a strong finite-size scaling dependence on both interaction strength and initial conditions (random, domain-wall, or Néel) as demonstrated by the data, which are well-described by fitting functions detailed in Tab. 3.
The long-time imbalance exhibits a strong finite-size scaling dependence on both interaction strength and initial conditions (random, domain-wall, or Néel) as demonstrated by the data, which are well-described by fitting functions detailed in Tab. 3.

The study meticulously dissects the intricacies of many-body localization, revealing how local resonances dictate the long-term evolution of imbalance measurements. This challenges conventional interpretations of finite-size scaling, suggesting that observed behaviors aren’t necessarily indicative of true many-body effects but rather artifacts of these resonances. It is akin to what Francis Bacon observed: “Knowledge is power,” and in this instance, a deeper knowledge of the system’s internal dynamics-specifically, these resonances-reveals the limitations of relying solely on macroscopic observations. The researchers don’t simply accept the apparent order; instead, they probe the underlying mechanisms, effectively reverse-engineering the system to understand the true nature of the localization process.

The Echo in the Static

The identification of local resonances as drivers of long-time imbalance dynamics in many-body localized systems does not resolve the inherent ambiguities, but rather reframes them. It suggests that the apparent ‘localization’ itself may be, in part, a consequence of observing a system perpetually echoing with its own internal structure. This raises a discomfiting question: to what degree is the observed absence of thermalization a genuine phase of matter, and to what degree a measurement artifact-a systematic misinterpretation of resonant behavior?

Future work must confront the challenge of disentangling these contributions. Finite-size scaling, previously assumed a reliable path toward the thermodynamic limit, now demands rigorous reevaluation. The subtle interplay between disorder, resonance frequencies, and measurement timescales necessitates novel analytical tools and simulation techniques. Perhaps the true signature of many-body localization lies not in the suppression of transport, but in the precise characterization of these internal echoes-their distribution, their decay, and their sensitivity to external perturbations.

The architecture of disorder, it seems, isn’t simply a random scattering of potential; it’s an instrument for generating complexity. Further investigations should explore whether similar resonant phenomena underpin other aspects of localization, and whether controlled manipulation of these resonances could reveal new pathways for quantum control-or, conversely, demonstrate the fundamental limits of predictability in strongly disordered systems.


Original article: https://arxiv.org/pdf/2601.05177.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/

See also:

2026-01-09 19:40