Pith. sign in

REVIEW 2 major objections 3 minor 88 references

The paper shows that in a strongly disordered spin chain, site-resolved autocorrelators reveal rare few-spin resonances that are hidden by the usual imbalance average, and that these resonances control the finite-size drift of the long-time

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 11:45 UTC pith:JR4RIM7A

load-bearing objection The site-resolved multi-peak result and the two-body peak formula are solid and worth knowing; the paper's quantitative Néel-state finite-size claim is not, and needs fixing. the 2 major comments →

arxiv 2601.05177 v1 pith:JR4RIM7A submitted 2026-01-08 cond-mat.dis-nn cond-mat.quant-gascond-mat.str-elquant-ph

Beyond the imbalance: site-resolved dynamics probing resonances in many-body localization

Asmi Haldar , Thibault Scoquart , Fabien Alet , Nicolas Laflorencie This is my paper
classification cond-mat.dis-nn cond-mat.quant-gascond-mat.str-elquant-ph
keywords many-body localizationimbalancesite-resolved autocorrelatorslocal resonancesrandom-field XXZ chainfinite-size scalingrare regionsquench dynamics
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper argues that the standard imbalance observable, averaged over sites, hides the most informative dynamics in a disordered many-body localized spin chain. Using exact numerics and an analytical few-site toy model, it shows that the distribution of site-resolved local autocorrelators develops secondary peaks whose positions are quantitatively explained by two- and three-spin resonances between nearby sites with nearly equal random fields. The authors establish that these rare local resonances control the sign and magnitude of the finite-size corrections to the long-time imbalance, and that the corrections depend on the initial state: approached from above for a Néel state, from below for random or domain-wall states. This matters because it turns a seemingly generic finite-size drift into a state-dependent, microscopic signature that can be probed with site-resolved imaging.

Core claim

Central claim: in the strongly disordered regime of the random-field XXZ spin chain, the long-time value of the imbalance is not a generic measure of localization but is shaped by rare local resonances. Most spins remain pinned, but a small density of sites — those where two or three neighboring random fields nearly coincide — undergo large oscillations and decorrelate from their initial value. The paper derives closed-form expressions for the positions of the corresponding peaks in the full-trace autocorrelator histogram (about 0.603 for two-body resonances and 0.46 for three-body resonances at the Heisenberg point, with explicit 1/L shifts), and shows they match exact diagonalization data.

What carries the argument

The carrying object is the local resonance: a pair (or triplet) of adjacent spins whose effective local fields are nearly equal, so the spins undergo a Rabi-like oscillation with a frequency set by the spin-flip coupling and the field difference δh. The paper models each such cluster in isolation, assuming neighboring spins are frozen and act only through effective fields; the two-site model gives a closed-form long-time autocorrelator G²/(1+G²), and averaging over all local spin configurations yields the peak positions given in Eqs. (3.10)–(3.11). This mechanism transfers directly to the imbalance: counting resonant versus frozen sites with probabilities proportional to 1/h and 1/h² produce

Load-bearing premise

The paper's quantitative predictions assume that spins neighboring a resonant cluster are perfectly frozen and that the numerical or experimental time window is effectively infinite; if long-lived, longer-range resonances or neighbor fluctuations act before the measured time, the predicted peak positions and 1/L slopes could shift.

What would settle it

Look for the predicted sign difference in the 1/L drift: measure the disorder-averaged long-time imbalance for Néel versus random initial states across several system sizes in the same platform, and check whether Néel approaches from above and random from below. A sharper test is to extend exact time evolution to t ≈ 10⁵ in an L ≈ 20 sample and verify that the two-body peak in P(A_j) remains at the analytical value; if it drifts or splits, the frozen-neighbor/infinite-time assumption fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • The long-time imbalance should approach its thermodynamic value from above for Néel initial states and from below for random or domain-wall states, with the finite-size coefficient set by the resonance abundance.
  • The secondary peaks in the distribution of site-resolved autocorrelators should be experimentally visible with roughly 20–30 initial states and time windows around a few hundred tunneling times.
  • A slow power-law-like decay of the imbalance seen in finite-size simulations of Néel quenches can be a transient crossover to a lower nonzero plateau, not evidence of delocalization in the thermodynamic limit.
  • Two-body resonance peaks occur with probability proportional to 1/h and three-body peaks proportional to 1/h²; increasing disorder shifts weight back to the main peak without moving the peak positions.
  • Free-fermion (Δ=0) results reproduce the same 1/L structure, so the mechanism is not interaction-specific.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Because the resonance statistics depend on the probability that nearby random fields are nearly equal, other disorder distributions (e.g., binary or Gaussian) should show weaker or stronger secondary peaks depending on the value of P(δh≈0); this is a testable extension the paper only hints at.
  • If the frozen-neighbor assumption fails at weaker disorder, the two-body peak should broaden and merge with the main peak; tracking that merging as h decreases could serve as a local probe of the approach to ergodicity.
  • The authors note that longer-range 'cat-state' resonances act on exponentially longer timescales. A natural extension is that those resonances will introduce additional, time-dependent shifts of the measured peak positions for times much larger than 10⁴, making the separation-of-timescales assumption testable in long-time quantum-simulator experiments.
  • In models without total-magnetization conservation, the combinatorial 1/L factors will differ or vanish, so the sign of the imbalance finite-size drift may flip or disappear; this suggests a way to distinguish conservation-law-driven effects from resonance effects.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 3 minor

Summary. The paper studies the random-field XXZ chain in the strongly disordered (MBL) regime and argues that the spatially averaged imbalance hides a rich site-resolved structure. Using Krylov time evolution and full diagonalization, the authors show that histograms of long-time, site-resolved autocorrelators display secondary peaks, which they attribute to two- and three-body local resonances. An analytic frozen-neighbor, few-site toy model yields parameter-free predictions for the peak positions and their 1/L shifts; these are compared with numerics for the free-fermion case (L=8–128) and the Heisenberg case (L=8–18). The paper then uses the statistics of these resonances to derive formulas for the finite-size scaling of the long-time imbalance for random, domain-wall, and Néel initial states, predicting negative 1/L corrections for generic random states and positive 1/L corrections for the Néel state. The authors also discuss finite-time and finite-sampling effects relevant to experiments.

Significance. If correct, the central claim—that local few-body resonances, invisible in the site-averaged imbalance, control the initial-state-dependent finite-size drift of the long-time imbalance—would be an important step toward reconciling conflicting numerical and experimental interpretations of imbalance dynamics in MBL. The paper has real strengths: the two-body resonance peak formula is an essentially parameter-free analytic derivation, and its agreement with exact numerics across a wide range of system sizes is impressive; the free-fermion comparison up to L=128 and the careful analysis of partial-trace convergence add credibility. The practical message that site-resolved and/or state-resolved measurements can expose physics hidden by spatial averaging is valuable and experimentally testable. However, the quantitative explanation of the Néel-state finite-size drift contains a specific derivation error (detailed below), and the infinite-time identification for some interactive data rests on an unverified timescale assumption. These issues do not invalidate the site-resolved peak analysis, but they do undermine one of the paper's headline conclusions as currently stated.

major comments (2)
  1. [Sec. IVC2c, Eqs. (4.9)-(4.10)] The derivation of the Néel finite-size correction is not valid as written. For a fixed Néel state every bond has M_{1-2}=0 and has opposite neighboring spins, so Table I selects a single cell whose delta_h -> 0 value is Delta^2/(1+Delta^2). The factors 1/4 and (1+1/L) in Eq. (4.9) come from Eqs. (4.2)-(4.3), which describe averaging over random initial states, not a fixed Néel configuration; the sentence 'assuming that averaging over disorder realizations provides the same combinatorial factors' does not supply the missing argument, since disorder averaging does not randomize the fixed local magnetization pattern. With the random-state value alpha/h ~ 0.40, the correct two-site expression gives a Néel intercept 1 - 0.5*0.40 ~ 0.80, whereas ED gives 0.717(7) and Eq. (4.10) gives 0.65. Thus no single alpha/h reproduces both the random and Néel fits, and the claimed quantitative explanation
  2. [Sec. IVB2 / Fig. 7(b) and Table III] The identification of time-averaged data at t_max ~ 10^4 with the infinite-time limit is an assumption, not a demonstrated convergence, for the interactive imbalance data used in Fig. 8 and Table III. The paper itself notes in Sec. IVB2 that exponentially slow cat-state resonances (Ref. [49]) exist and cannot be captured by the few-site model. If such rare long-range resonances decorrelate sites on timescales beyond 10^4, the fitted 1/L coefficients in Table III could shift. Please compare the time-averaged Krylov results against exact infinite-time ED values for the same system sizes and initial states (available for L <= 18), or show explicitly that the fitted coefficients are stable when the time window is varied.
minor comments (3)
  1. [Eq. (4.6) and App. S3] The identification alpha/h = 2 exp(-1/xi_eff) appears inconsistent with the free-fermion expression in App. S3, where g(r) ~ q^r with q = exp(-2/xi_eff), giving alpha/h = 2q = 2 exp(-2/xi_eff). Please check the exponent and the line '1/xi_eff = ln(h/h0)'.
  2. [General notation] Several typographical issues should be corrected: 'App. S23' should be 'App. S2.3'; the accent in 'INeel' is garbled; and Table III uses a mixture of L, L-1, and L-(1+(-1)^L)/2 denominators, which should be unified for readability.
  3. [Fig. 7(a)] In Fig. 7(a) the featurure at the predicted two-body value is described as a 'shoulder' for small N_s but a 'rounded secondary peak' for larger N_s. It would be helpful to state explicitly whether the location of this feature is fitted or fixed to the analytic value, since the gradual build-up is an important experimental claim.

Circularity Check

2 steps flagged

Multi-peak toy-model analysis is self-contained, but the Sec. IVC imbalance 'predictions' are one-parameter fits (alpha/h) and the Neel 1/L coefficient is imported from random-state combinatorics rather than derived.

specific steps
  1. fitted input called prediction [Sec. IVC1, Eq. (4.1) and Eq. (4.5); Fig. 8 / Tab. III]
    "Iavg(L) ≈ (1−α/h) + Zfluct.(∞) α/h + O(1/h²) ... where α is an O(1) non-universal number. ... Reinjecting this expression into the average imbalance Eq. (4.1), we can rewrite it as IRand_avg(L)≈1−xΔ α/h −xΔ α/h 1/(L−1), xΔ=(Δ²+2)/(4(Δ²+1)). (4.5)"

    alpha/h is not fixed by the toy model; it is a free parameter inferred from the same kind of numerical data (histogram weights in Fig. 6, and equivalently the intercept of the imbalance fits in Fig. 8/Tab. III). Because the same alpha/h sets both the L→∞ intercept and the 1/L slope in Eq. (4.5), the slope is a deterministic function of the fitted intercept: b = -xDelta/(1-xDelta) * (1 - intercept). Thus the quantitative 'prediction' of the negative 1/L correction is a one-parameter reparametrization of the fit, not an independent derivation. The sign of 1/L is still a genuine derived feature, so this is partial, not total, circularity.

  2. other [Sec. IVC2c, Eq. (4.9)-(4.10)]
    "For this state, all two-site local configuration have M1−2=0 and furthermore are neighbored by spins with opposite direction. This means that only one contribution out of the four possible in Tab. I is present for the Néel state. Repeating the computation and assuming that averaging over disorder realizations provides the same combinatorial factors that from averaging over initial states, we expect Zfluct(∞)≈ 1/4 Δ²/(Δ²+1)(1+1/L), (4.9)"

    For a fixed Néel state, the preceding sentence states that only one cell of Tab. I contributes; the delta-h→0 value of that single cell is Delta^2/(1+Delta^2), with no factor 1/4 and no 1+1/L. The factors 1/4 and 1+1/L are the random-initial-state probabilities of Eqs. (4.2)-(4.3). Inserting them into the Néel computation imports the initial-state averaging that a fixed Néel state does not possess. Consequently the positive 1/L term in Eq. (4.10) is not a derived consequence of the toy model for the Néel state; it reproduces the positive slope fitted from ED data by borrowing random-state combinatorics, making that 'prediction' effectively a post-hoc accommodation rather than a derivation.

full rationale

The central toy-model analysis of the secondary peaks is self-contained and independently checked: Eqs. (3.7)-(3.11) with Tab. I give closed-form peak positions from a two/three-site effective Hamiltonian with frozen neighbors, and the ED/free-fermion comparisons in Figs. 4-5 are genuine numerical benchmarks. The 1/h and 1/h^2 resonance probabilities follow directly from the box distribution and are separately verified in Fig. 6. No load-bearing self-citation chain or uniqueness theorem is used; self-citations (e.g. Refs. [47-49]) are contextual or flag limitations. The Sec. IVC imbalance predictions, however, are not parameter-free: Eq. (4.1) introduces alpha/h as an unspecified non-universal number, and Eqs. (4.5), (4.8), (4.10) all inherit it. Because the same alpha/h controls both intercept and 1/L slope, the agreement in Fig. 8 is a one-parameter consistency check. The Néel case is worse: Eq. (4.9) explicitly borrows random-state combinatorial factors even though the text says only one cell of Tab. I contributes. Thus the positive 1/L correction for Néel is not derived from the frozen-neighbor toy model. The paper itself acknowledges finite-time and long-range cat-state limitations (Sec. IVB2, Ref. [49]), which are not circular but further cap the force of the finite-size claims. Overall, the multi-peak resonance claims are honest and independent, but the quantitative imbalance 'explanations' are partially fit-backed, so the circularity score is 6 rather than 0.

Axiom & Free-Parameter Ledger

2 free parameters · 5 axioms · 0 invented entities

The central analytical peak positions are derived from a minimal few-site model with no new physical entities and no fitted constants. The main free parameter is the resonance probability prefactor α/h used to translate the toy model into quantitative predictions for the disorder-averaged imbalance. The frozen-neighbor and finite-time/infinite-time identifications are explicit modeling assumptions in the paper.

free parameters (2)
  • α (2-body resonance probability prefactor) = α/h ≈ 2.9 h^{-0.99} from Fig. 6(b); α appears in Eqs. (4.1), (4.5), (4.8), (4.10)
    The probability that a site participates in a 2-body resonance is parametrized as α/h at strong disorder. The prefactor α is not derived from the Hamiltonian but extracted from fits to the numerical histograms and imbalance data.
  • 3-body resonance weight prefactor = ≈ 0.0089 h^{-2.07} from Fig. 6(b)
    The integrated weight of the tertiary 3-body peak is fitted to a power law in h. This is used to support the 1/h^2 scaling estimate but is not needed for the main peak-position predictions.
axioms (5)
  • standard math Standard unitary quantum mechanics and Pauli algebra for spin-1/2 chains
    The time evolution, autocorrelators, and eigenstate decompositions use standard quantum mechanical identities.
  • domain assumption The random-field XXZ Hamiltonian with box disorder [−h,h] is in the MBL phase at h=10, Δ=1
    The paper relies on established static MBL indicators (spectral statistics, eigenstate entanglement) from refs. [47-52] to place the parameters in the MBL regime.
  • ad hoc to paper Frozen-neighbor (product-state background) approximation for resonant clusters
    The 2- and 3-site toy models (Eqs. (3.2)-(3.3), App. S2) assume that spins neighboring a resonant cluster are pinned and do not fluctuate. The authors explicitly state this condition in Sec. IVA.
  • ad hoc to paper Time-averaged data at t≈5×10^3–10^4 represent the infinite-time limit
    Figs. 3, 7, and 8 use time-averaged data at finite t as proxies for t=∞. The authors acknowledge that exponentially slow, longer-range cat-state resonances exist (Sec. IVB2) and cannot be captured by the toy model, so this time-scale separation is an assumption.
  • standard math Total magnetization conservation and the combinatorial counts in the zero-magnetization sector
    The counting of D0 and Dp in Eqs. (3.9) and App. S2 uses the constraint S^z_tot=0 and standard combinatorial identities.

pith-pipeline@v1.3.0-alltime-deepseek · 32760 in / 19349 out tokens · 203916 ms · 2026-08-03T11:45:32.315428+00:00 · methodology

0 comments
read the original abstract

We explore the limitations of using imbalance dynamics as a diagnostic tool for many-body localization (MBL) and show that spatial averaging can mask important microscopic features. Focusing on the strongly disordered regime of the random-field XXZ chain, we use state-of-the-art numerical techniques (Krylov time evolution and full diagonalization) to demonstrate that site-resolved spin autocorrelators reveal a rich and complex dynamical behavior that is obscured by the imbalance observable. By analyzing the time evolution and infinite-time limits of these local probes, we reveal resonant structures and rare local instabilities within the MBL phase. These numerical findings are supported by an analytical, few-site toy model that captures the emergence of a multiple-peak structure in local magnetization histograms, which is a hallmark of local resonances. These few-body local effects provide a more detailed understanding of ergodicity-breaking dynamics, and also allow us to explain the finite-size effects of long-time imbalance, and its sensitivity to the initial conditions in quench protocols. Overall, our experimentally testable predictions highlight the necessity of a refined, site-resolved approach to fully understand the complexities of MBL and its connection to rare-region effects.

Figures

Figures reproduced from arXiv: 2601.05177 by Asmi Haldar, Fabien Alet, Nicolas Laflorencie, Thibault Scoquart.

Figure 1
Figure 1. Figure 1: Illustration of what spatial averaging of imbalance hides — Local spin dynamics for two typical (𝐿 = 24 sites) samples, at strong disorder (ℎ = 10) and interaction Δ = 1. Starting from an initial product state |𝑚⟩ (shown on each plots), the system follows unitary time evolution Eq. (2.1). The main panels (a-b) show the (time-averaged) site-resolved spin autocorrelators [𝑍 (𝑚) 𝑗 (𝑡)]𝑇, see Eq. (2.3), for tw… view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of what initial-state averaging hides — Time￾averaged autocorrelators of a single 𝐿 = 20 disordered sample having the same first 20 random fields as the one in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Histograms of the various quantities entering in the average [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Histograms of the (infinite-time, site-resolved) autocorrela [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Justification of the toy-model description. Left: His￾tograms of the full traces AFull 𝑗 computed using full ED at very strong disorder (ℎ ∈ [10, 100]), for system size 𝐿 = 14, averaged over 𝑁𝑟 ∼ 105 disorder realizations. As ℎ increases, the dominant peak at A ≃ 1 grows, while the secondary two-body (2br) and three-body (3br) resonance peaks progressively lose weight. The two vertical dotted lines separat… view at source ↗
Figure 8
Figure 8. Figure 8: Finite-size scaling of the long-time imbalance plotted [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

88 extracted references · 4 linked inside Pith

  1. [1]

    5(b) where full-trace ED data for𝐿 = 18 sites are shown

    Heisenberg case Δ= 1 We start with the interacting caseΔ= 1 in Fig. 5(b) where full-trace ED data for𝐿 = 18 sites are shown. The analytical predictionsfor2-and3-bodyresonances(fromTab.II)nicely coincide with the positions of the peaks in the numerically obtained histogram. More precisely, the inset (ii) shows how the peaks positions shift with1/𝐿, in very...

  2. [2]

    infinite temperature

    in the initial state is0 or not. For the polarized states|↑↑⟩;|↓↓⟩ withM1−2=±1,theautocorrelatorstaysconstant 𝑍±1 1,2(𝑡)≈ 1, ∀𝑡. For initial states with vanishing magnetizationM1−2 = 0 {|↑↓⟩;|↓↑⟩} on the other hand, the autocorrelators follow an oscillatory behavior: 𝑍0 1,2(𝑡)= 1− 2 1+G 2 sin2(Ω𝑡). (3.4) withΩ,G determined by the interactionΔ, the differe...

  3. [3]

    However,performingthefulltracecanquicklybeunattainable as the number of states𝑁H becomes too large (in practice it startstobenumericallycostlyfor 𝐿 ≳ 20)

    Non-interacting case Δ= 0: free-fermion results For non-interacting (Δ= 0) systems, a wider range of sizes areavailable( 𝐿= 8−128)thankstofree-fermioncalculations. However,performingthefulltracecanquicklybeunattainable as the number of states𝑁H becomes too large (in practice it startstobenumericallycostlyfor 𝐿 ≳ 20). Wefirstcharacter- ize if it is enough ...

  4. [4]

    Finite number of initial states Exceptforverysmallsystems,experimentalconstraintspre- vent performing the full trace required to obtain the autocor- relatorAfull 𝑗 , leaving only the partial estimatorApartial 𝑗 acces- sible. Arelevantquestionisthus: howmanyinitialstates(per realizationofdisorder)areneededsothat Apartial 𝑗 capturesthe mainfeaturesofAfull 𝑗...

  5. [5]

    Finite time in simulations/experiments Another important experimental aspect is that the system canstayfullyisolatedfromtheenvironmentonlyforalimited amount of time (ranging up to a few hundreds-thousands of typicalhopping/spin-fliptimes). Wethusneedtodetermineat 10 what typical times the influence of the few-body short-range resonances (which we argue pr...

  6. [6]

    Within this strong disorder description, justified by data shown in Fig

    Statistical modeling at strong disorder Assuming a simplified statistical description of a strongly disordered chain (ℎ≫ 1) where spins either belong to a few- site resonant region or are almost totally frozen⟨𝜎𝑧⟩≃± 1, we expect the average infinite-time imbalance to be given by Iavg(𝐿) ≈ 1 𝐿 ©­­­ « ∑︁ frozen sites O( 1)+ ∑︁ fluctuating sites 𝑍𝑗(∞) ª®®® ¬...

  7. [7]

    Random initial states — We first consider the dy- namics from a single random initial state|𝑚⟩ for each disor- dered sample, taking the average afterwards

    Initial state dependence a. Random initial states — We first consider the dy- namics from a single random initial state|𝑚⟩ for each disor- dered sample, taking the average afterwards. From the local- resonance toy model and for a pair of neighboring sites with similar fields (𝛿ℎ≃ 0), we expect qualitatively different dy- namics if|𝑚⟩ has corresponding pol...

  8. [8]

    8) for quench dynamics in experiments and numerical simulations

    Discussion We now briefly discuss the consequences of these results (in particular Fig. 8) for quench dynamics in experiments and numerical simulations. Compared to a generic random initial state,theNéelconfigurationisknowntobehighlypeculiar: all its bonds are of the form↑↓or↓↑, and are therefore flippable under the XXZ Hamiltonian. In other words, in the...

    2025

  9. [9]

    Krylov evolution TimeevolutionusingKrylovspaceisanefficientnumericaltechniqueproposedbyNautsandWyatt[75]toapproximatetheac- tionofthetime-evolutionoperator 𝑒−𝑖𝐻𝑡|𝜓0⟩,where 𝐻isaHermitianHamiltonianactingonahigh-dimensionalHilbertspaceand |𝜓0⟩isaninitialstate. Ratherthancomputingthefullmatrixexponential,whichiscomputationallyinfeasibleforlargemany-body 14 s...

  10. [10]

    (S1) Using the discrete version of the above equation, we exemplify the procedure in Fig

    Time averaged local magnetization In order to smooth the curves and get rid of fast oscillations, we performed a time-average of the autocorrelator data [𝑍𝑚 𝑗(𝑡)]𝑇 = 1 𝑡−𝑡min ∫ 𝑡 𝑡min 𝑍𝑚 𝑗(𝜏)d𝜏. (S1) Using the discrete version of the above equation, we exemplify the procedure in Fig. S1, which shows three representative behaviorsthatappearinsample1(Fig.1(...

  11. [11]

    toy-state

    2-site resonances We first consider a toy model that consists of 2 sites embedded in a trivially localized product-state background. We describe this by the following "toy-state" ansatz |TS⟩=|↑↑↓...⟩⊗|◦ 1◦2⟩⊗|↓↑↓ ...⟩, (S1) where|◦1◦2⟩ is the pair of (potentially fluctuating) sites for which we want to characterize the quantum dynamics. Assuming that the ...

  12. [12]

    S2(a), along with the bond variable𝛿ℎ𝑗 = ℎ𝑗+1−ℎ𝑗, which represents the field difference between nearest neighbors

    Numerical checks of 2-body resonances Two-body resonances can be numerically observed using another𝐿= 20 sample, whose random field configuration is shown in Fig. S2(a), along with the bond variable𝛿ℎ𝑗 = ℎ𝑗+1−ℎ𝑗, which represents the field difference between nearest neighbors. We focus on sites𝑗 = 9 and 10, which exhibit a very small bond variable|𝛿ℎ|≈ 0....

  13. [13]

    infinite temperature

    Three-body resonances One can use a similar approach for 3 consecutive sites in a locally uniform region. The analytical calculation is still possible butabitlong,sowewillonlyquotethemainresultshere. Inordertodealwithsimpleexpressions,weassumethatlocalrandom fields are equal (ℎ1=ℎ2=ℎ3=ℎ). The 3-site effective Hamiltonian is H(3) eff = 2∑︁ 𝑗=1 𝜎𝑥 𝑗𝜎𝑥 𝑗+1+𝜎...

  14. [14]

    Rev.109, 1492 (1958)

    P.W.Anderson,Absenceofdiffusionincertainrandomlattices, Phys. Rev.109, 1492 (1958)

    1958

  15. [15]

    Billy, V

    J. Billy, V. Josse, Z. Zuo, A. Bernard, B. Hambrecht, P. Lugan, D. Clément, L. Sanchez-Palencia, P. Bouyer, and A. Aspect, Direct observation of anderson localization of matter waves in a controlled disorder, Nature453, 891 (2008)

    2008

  16. [16]

    Roati, C

    G. Roati, C. D’Errico, L. Fallani, M. Fattori, C. Fort, M. Zac- canti, G. Modugno, M. Modugno, and M. Inguscio, Anderson localization of a non-interacting bose–einstein condensate, Na- ture453, 895 (2008)

    2008

  17. [17]

    H. Hu, A. Strybulevych, J. H. Page, S. E. Skipetrov, and B. A. vanTiggelen,Localizationofultrasoundinathree-dimensional elastic network, Nature Physics4, 945 (2008)

    2008

  18. [18]

    Lemarié, H

    G. Lemarié, H. Lignier, D. Delande, P. Szriftgiser, and J. C. Garreau, Critical State of the Anderson Transition: Between a Metal and an Insulator, Phys. Rev. Lett.105, 090601 (2010), publisher: American Physical Society

    2010

  19. [19]

    D.A.Huse,R.Nandkishore,andV.Oganesyan,Phenomenology of fully many-body-localized systems, Phys. Rev. B90, 174202 (2014)

    2014

  20. [20]

    M.Serbyn,Z.Papić,andD.A.Abanin,Localconservationlaws and the structure of the many-body localized states, Phys. Rev. Lett.111, 127201 (2013)

    2013

  21. [21]

    V. Ros, M. Müller, and A. Scardicchio, Integrals of motion in the many-body localized phase, Nucl. Phys. B891, 420 (2015)

    2015

  22. [22]

    J. Z. Imbrie, On many-body localization for quantum spin chains, Journal of Statistical Physics163, 998–1048 (2016)

    2016

  23. [23]

    J. H. Bardarson, F. Pollmann, and J. E. Moore, Unbounded growth of entanglement in models of many-body localization, Phys. Rev. Lett.109, 017202 (2012)

    2012

  24. [24]

    Vosk and E

    R. Vosk and E. Altman, Many-body localization in one dimen- sion as a dynamical renormalization group fixed point, Phys. Rev. Lett.110, 067204 (2013)

    2013

  25. [25]

    Serbyn, Z

    M. Serbyn, Z. Papić, and D. A. Abanin, Quantum quenches in themany-bodylocalizedphase,Phys.Rev.B 90,174302(2014)

    2014

  26. [26]

    Lukin, M

    A. Lukin, M. Rispoli, R. Schittko, M. E. Tai, A. M. Kauf- man,S.Choi,V.Khemani,J.Léonard,andM.Greiner,Probing entanglement in a many-body-localized system, Science364, 256–260 (2019)

    2019

  27. [27]

    Schreiber, S

    M. Schreiber, S. S. Hodgman, P. Bordia, H. P. Lüschen, M. H. Fischer, R. Vosk, E. Altman, U. Schneider, and I. Bloch, Ob- servation of many-body localization of interacting fermions in a quasirandom optical lattice, Science349, 842 (2015)

    2015

  28. [28]

    Smith, A

    J. Smith, A. Lee, P. Richerme, B. Neyenhuis, P. W. Hess, 21 P. Hauke, M. Heyl, D. A. Huse, and C. Monroe, Many-body localizationinaquantumsimulatorwithprogrammablerandom disorder, Nature Physics12, 907 (2016)

    2016

  29. [29]

    J.-Y.Choi,S.Hild,J.Zeiher,P.Schauß,A.Rubio-Abadal,T.Yef- sah,V.Khemani,D.A.Huse,I.Bloch,andC.Gross,Exploring the many-body localization transition in two dimensions, Sci- ence352, 1547 (2016)

    2016

  30. [30]

    Kohlert, S

    T. Kohlert, S. Scherg, X. Li, H. P. Lüschen, S. Das Sarma, I. Bloch, and M. Aidelsburger, Observation of many-body lo- calization in a one-dimensional system with a single-particle mobility edge, Phys. Rev. Lett.122, 170403 (2019)

    2019

  31. [31]

    Chiaroet al., Direct measurement of nonlocal interactions in the many-body localized phase, Phys

    B. Chiaroet al., Direct measurement of nonlocal interactions in the many-body localized phase, Phys. Rev. Res.4, 013148 (2022)

    2022

  32. [32]

    Schneider, Coupling identical one-dimensional many- body localized systems, Phys

    P.Bordia,H.P.Lüschen,S.S.Hodgman,M.Schreiber,I.Bloch, and U. Schneider, Coupling identical one-dimensional many- body localized systems, Phys. Rev. Lett.116, 140401 (2016)

    2016

  33. [33]

    H. P. Lüschen, P. Bordia, S. Scherg, F. Alet, E. Altman, U.Schneider,andI.Bloch,ObservationofSlowDynamicsnear the Many-Body Localization Transition in One-Dimensional Quasiperiodic Systems, Phys. Rev. Lett.119, 260401 (2017)

    2017

  34. [34]

    Yaoet al., Observation of many-body fock space dynamics in two dimensions, Nature Physics19, 14595 (2023)

    Y. Yaoet al., Observation of many-body fock space dynamics in two dimensions, Nature Physics19, 14595 (2023)

    2023

  35. [35]

    Li et al., Many-body delocalization with a two- dimensional 70-qubit superconducting quantum simulator, arXiv:2507.16882

    T.-M. Li et al., Many-body delocalization with a two- dimensional 70-qubit superconducting quantum simulator, arXiv:2507.16882

    Pith/arXiv arXiv

  36. [36]

    Yu,A.Chan,T.Wahl,andJ.yoonChoi,Stabilityofmany-body localization in two dimensions, arXiv:2508.20699

    J.Hur,J.Li,B.Lee,K.Kwon,M.Kim,S.Hwang,S.Kim,Y.S. Yu,A.Chan,T.Wahl,andJ.yoonChoi,Stabilityofmany-body localization in two dimensions, arXiv:2508.20699

    Pith/arXiv arXiv

  37. [37]

    A.Lunkin etal.,Evidenceforatwo-dimensionalquantumglass state at high temperatures (2026), arXiv:2601.01309

    Pith/arXiv arXiv 2026

  38. [38]

    B93, 060201 (2016)

    D.J.Luitz,N.Laflorencie,andF.Alet,Extendedslowdynamical regimeclosetothemany-bodylocalizationtransition,Phys.Rev. B93, 060201 (2016)

    2016

  39. [39]

    E. J. Torres-Herrera, A. M. García-García, and L. F. Santos, Generic dynamical features of quenched interacting quantum systems: Survival probability, density imbalance, and out-of- time-ordered correlator, Phys. Rev. B97, 060303 (2018)

    2018

  40. [40]

    E. V. H. Doggen, F. Schindler, K. S. Tikhonov, A. D. Mirlin, T.Neupert, D.G.Polyakov,andI.V.Gornyi,Many-bodylocal- ization and delocalization in large quantum chains, Phys. Rev. B98, 174202 (2018)

    2018

  41. [41]

    Chanda, P

    T. Chanda, P. Sierant, and J. Zakrzewski, Time dynamics with matrix product states: Many-body localization transition of large systems revisited, Phys. Rev. B101, 035148 (2020)

    2020

  42. [42]

    Benini, P

    L. Benini, P. Naldesi, R. A. Römer, and T. Roscilde, Loschmidt echo singularities as dynamical signatures of strongly localized phases, New Journal of Physics23, 023030 (2021)

    2021

  43. [43]

    Nandy, F

    S. Nandy, F. Evers, and S. Bera, Dephasing in strongly dis- ordered interacting quantum wires, Phys. Rev. B103, 085105 (2021)

    2021

  44. [44]

    P.SierantandJ.Zakrzewski,Challengestoobservationofmany- body localization, Phys. Rev. B105, 224203 (2022)

    2022

  45. [45]

    Prasad and A

    Y. Prasad and A. Garg, Initial state dependent dynamics across themany-bodylocalizationtransition,Phys.Rev.B 105,214202 (2022)

    2022

  46. [46]

    Scoquart, I

    T. Scoquart, I. V. Gornyi, and A. D. Mirlin, Scaling of many- body localization transitions: Quantum dynamics in fock space and real space, Phys. Rev. B112, 064203 (2025)

    2025

  47. [47]

    Agarwal, S

    K. Agarwal, S. Gopalakrishnan, M. Knap, M. Müller, and E. Demler, Anomalous diffusion and griffiths effects near the many-bodylocalizationtransition,Phys.Rev.Lett. 114,160401 (2015)

    2015

  48. [48]

    T. L. M. Lezama, S. Bera, and J. H. Bardarson, Apparent slow dynamics in the ergodic phase of a driven many-body localized systemwithoutextensiveconservedquantities,Phys.Rev.B 99, 161106 (2019)

    2019

  49. [49]

    D. M. Long, P. J. D. Crowley, V. Khemani, and A. Chandran, Phenomenologyoftheprethermalmany-bodylocalizedregime, Phys. Rev. Lett.131, 106301 (2023)

    2023

  50. [50]

    Haldar, Kohlrausch regime of slow relaxation and its phe- nomenology in isolated quantum many-body systems with strong disorder, Phys

    A. Haldar, Kohlrausch regime of slow relaxation and its phe- nomenology in isolated quantum many-body systems with strong disorder, Phys. Rev. B111, 075113 (2025)

    2025

  51. [51]

    Nandkishore and D

    R. Nandkishore and D. A. Huse, Many-body localization and thermalizationinquantumstatisticalmechanics,AnnualReview of Condensed Matter Physics6, 15–38 (2015)

    2015

  52. [52]

    D.A.Abanin,E.Altman,I.Bloch,andM.Serbyn,Colloquium: Many-bodylocalization,thermalization,andentanglement,Rev. Mod. Phys.91, 021001 (2019)

    2019

  53. [53]

    T.Szołdra,P.Sierant,M.Lewenstein,andJ.Zakrzewski,Track- ing locality in the time evolution of disordered systems, Phys. Rev. B107, 054204 (2023)

    2023

  54. [54]

    Vallejo-Fabila, A

    I. Vallejo-Fabila, A. K. Das, S. Choudhury, and L. F. San- tos,Single-sitemeasurementsasprobesofmany-bodyquantum chaos, Phys. Rev. E112, 044208 (2025)

    2025

  55. [55]

    B111, L220202 (2025)

    P.Brighi,M.Ljubotina,andM.Serbyn,Probingthemany-body localizedspin-glassphasethroughquenchdynamics,Phys.Rev. B111, L220202 (2025)

    2025

  56. [56]

    De Roeck and F

    W. De Roeck and F. m. c. Huveneers, Stability and instability towardsdelocalizationinmany-bodylocalizationsystems,Phys. Rev. B95, 155129 (2017)

    2017

  57. [57]

    Thiery, F

    T. Thiery, F. m. c. Huveneers, M. Müller, and W. De Roeck, Many-body delocalization as a quantum avalanche, Phys. Rev. Lett.121, 140601 (2018)

    2018

  58. [58]

    W. S. Bakr, J. I. Gillen, A. Peng, S. Fölling, and M. Greiner, A quantumgasmicroscopefordetectingsingleatomsinahubbard- regime optical lattice, Nature462, 74 (2009)

    2009

  59. [59]

    M. F. Parsons, F. Huber, A. Mazurenko, C. S. Chiu, W. Seti- awan,K.Wooley-Brown,S.Blatt,andM.Greiner,Site-resolved imaging of fermionic6Li in an optical lattice, Phys. Rev. Lett. 114, 213002 (2015)

    2015

  60. [60]

    J.Colbois,F.Alet,andN.Laflorencie,Interaction-DrivenInsta- bilities in the Random-Field XXZ Chain, Phys. Rev. Lett.133, 116502 (2024)

    2024

  61. [61]

    Colbois, F

    J. Colbois, F. Alet, and N. Laflorencie, Statistics of systemwide correlationsintherandom-fieldXXZchain: Importanceofrare events in the many-body localized phase, Phys. Rev. B110, 214210 (2024)

    2024

  62. [62]

    B112, 224207 (2025)

    N.Laflorencie,J.Colbois,andF.Alet,Catstatescarryinglong- rangecorrelationsinthemany-bodylocalizedphase,Phys.Rev. B112, 224207 (2025)

    2025

  63. [63]

    A.PalandD.A.Huse,Many-bodylocalizationphasetransition, Phys. Rev. B82, 174411 (2010)

    2010

  64. [64]

    D.J.Luitz,N.Laflorencie,andF.Alet,Many-bodylocalization edge in the random-field Heisenberg chain, Phys. Rev. B91, 081103(R) (2015)

    2015

  65. [65]

    Sierant, M

    P. Sierant, M. Lewenstein, and J. Zakrzewski, Polynomially fil- teredexactdiagonalizationapproachtomany-bodylocalization, Phys. Rev. Lett.125, 156601 (2020)

    2020

  66. [66]

    In our case, we employ pe- riodic boundary conditions, for which no such restriction is necessary

    When open boundary conditions are used, it is common to ex- clude a few sites near the open ends, since those sites are not representative of the bulk physics. In our case, we employ pe- riodic boundary conditions, for which no such restriction is necessary

  67. [67]

    S. P. Lim and D. N. Sheng, Many-body localization and transi- tion by density matrix renormalization group and exact diago- nalization studies, Phys. Rev. B94, 045111 (2016)

    2016

  68. [68]

    Laflorencie, G

    N. Laflorencie, G. Lemarié, and N. Macé, Chain breaking and 22 Kosterlitz-Thouless scaling at the many-body localization tran- sition in the random-field Heisenberg spin chain, Phys. Rev. Research2, 042033(R) (2020)

    2020

  69. [69]

    Roy and D

    S. Roy and D. E. Logan, Fock-space anatomy of eigenstates across the many-body localization transition, Phys. Rev. B104, 174201 (2021)

    2021

  70. [70]

    If𝐿 is odd, we can arbitrarily choose to take(𝐿+ 1)/2 spins↑ and (𝐿− 1)/2 spins↓

    For even𝐿, initial states have𝐿/2spins↑and𝐿/2spins↓. If𝐿 is odd, we can arbitrarily choose to take(𝐿+ 1)/2 spins↑ and (𝐿− 1)/2 spins↓

  71. [71]

    Colbois and N

    J. Colbois and N. Laflorencie, Breaking the chains: Extreme value statistics and localization in random spin chains, Phys. Rev. B108, 144206 (2023)

    2023

  72. [72]

    Karch, S

    S. Karch, S. Bandyopadhyay, Z.-H. Sun, A. Impertro, S. Huh, I. P. Rodríguez, J. F. Wienand, W. Ketterle, M. Heyl, A.Polkovnikov,I.Bloch,andM.Aidelsburger,Probingquantum many-body dynamics using subsystem loschmidt echos (2025), arXiv:2501.16995

    Pith/arXiv arXiv 2025

  73. [73]

    Yefsah, Quantum gas microscopy of fermions in the continuum, Phys

    T.deJongh,J.Verstraten,M.Dixmerias,C.Daix,B.Peaudecerf, and T. Yefsah, Quantum gas microscopy of fermions in the continuum, Phys. Rev. Lett.134, 183403 (2025)

    2025

  74. [74]

    [31], observingtheimbalancedecayfrom 0.73to 0.72wouldrequire a time of at least𝑡≈ 1010

    Assuming an exponent𝛽 = 0.0003 as reported in Ref. [31], observingtheimbalancedecayfrom 0.73to 0.72wouldrequire a time of at least𝑡≈ 1010

  75. [75]

    S. A. Weidinger, S. Gopalakrishnan, and M. Knap, Self- consistent hartree-fock approach to many-body localization, Phys. Rev. B98, 224205 (2018)

    2018

  76. [76]

    Hauschild, F

    J. Hauschild, F. Heidrich-Meisner, and F. Pollmann, Domain- wall melting as a probe of many-body localization, Phys. Rev. B94, 161109 (2016)

    2016

  77. [77]

    Wanget al., Exploring hilbert-space fragmentation on a superconducting processor, PRX Quantum6, 010325 (2025)

    Y.-Y. Wanget al., Exploring hilbert-space fragmentation on a superconducting processor, PRX Quantum6, 010325 (2025)

    2025

  78. [78]

    Humpert, D

    L. Humpert, D. M. Kennes, and J.-N. Herre, Tree ten- sor networks for many-body localization in two dimensions, arXiv:2512.19389

    arXiv

  79. [79]

    G.Biroli,A.K.Hartmann,andM.Tarzia,Large-deviationanal- ysisofrareresonancesforthemany-bodylocalizationtransition, Phys. Rev. B110, 014205 (2024)

    2024

  80. [80]

    Scoquart, I

    T. Scoquart, I. V. Gornyi, and A. D. Mirlin, Role of fock- spacecorrelationsinmany-bodylocalization,Phys.Rev.B 109, 214203 (2024)

    2024

Showing first 80 references.