Statistical levels and spatial modes of Fock-space heterogeneity in many-body localization crossovers

arXiv:2609.15323 · cond-mat.dis-nn, cond-mat.stat-mech · Submitted 2026-09-14 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.

Kai: We're looking at the paper: "Statistical levels and spatial modes of Fock-space heterogeneity in many-body localization crossovers".

Mira: Here's the gist of the paper. Statistical levels and spatial modes of Fock-space heterogeneity in many-body localization crossovers is a statistical framework based on configuration-space distance distributions that uses a variance decomposition to separate fluctuations within eigenstates, between eigenstates of one sample, and between samples.

Kai: At the exact-diagonalization sizes studied, the clearest difference occurs among disorder realizations or quasiperiodic phase samples, whereas within-eigenstate and within-sample eigenstate-to-eigenstate contributions remain broadly comparable.

Mira: Spatial covariances show that random outer fluctuations have a much stronger uniform component, whereas quasiperiodic phase fluctuations are organized more strongly at finite wave number and partly cancel in the spatial average controlling the distance center.

Kai: Pruning redundant reasoning tokens The framework connects Fock-space probability structure to local observables in real space by using a probability-weighted all-pairs distance statistic that groups pairs of configurations drawn independently from the same eigenstate’s configuration probability distribution by their separation to form a configuration-space distance distribution.

Paper summary: Kai: So this paper, "Statistical levels and spatial modes of Fock-space heterogeneity in many-body localization crossovers," it’s basically setting up a statistical way to look at how different the probability structures are when you're moving between different types of many-body localization.

Mira: Right. The core idea is using configuration space distance distributions and a variance decomposition to split the fluctuations into three parts: stuff inside one eigenstate, differences between eigenstates of one sample, and differences between samples themselves.

Lev: So it’s trying to separate the noise sources in these complex systems by looking at how far apart different configurations are in configuration space.

Kai: And they applied this to random-cosine and quasiperiodic-cosine Ising ensembles that all share the same one-site field marginal, and what they found is that the clearest difference between those two types of disorder or phase samples happens at the outer level.

Mira: That means while inside a single eigenstate or between eigenstates of one sample, things are pretty comparable across those different disorder realizations.

Lev: But when you look at the differences between the overall samples themselves, that’s where you see a sharper distinction in how they behave during this crossover.

Kai: They also looked at spatial organization—how those outer fluctuations are spread out across space—and they found that random outer fluctuations have a much stronger uniform component compared to the quasiperiodic phase fluctuations.

Mira: And for the quasiperiodic ones, even though they might look more organized at some wave number, they partly cancel out when you take the average over all of space, which affects the center of that distance distribution.

Conclusion: Kai: So to wrap up this paper on "Statistical levels and spatial modes of Fock-space heterogeneity in many-body localization crossovers," it’s about using configuration space geometry to distinguish where the noise is coming from in these systems.

Mira: The main thing they show is that the internal fluctuations—within a single eigenstate or within one sample—stay pretty similar regardless of whether you have random or quasiperiodic disorder.

Lev: But when you look at the differences between different disorder realizations, those are where the real physical signal lies in this study.

Kai: The title points to statistical levels and spatial modes, so they’re using a variance decomposition to separate the variation into these distinct levels of fluctuation.

Mira: And their finding is that while state-to-state variations are broadly comparable, the outer ensemble variation carries decisive physical information about the system type.

Lev: It suggests that even when you’re comparing two different physical ensembles, like random versus quasiperiodic, the difference in how they behave across samples is a key feature.

Kai: This work connects the probability structure of Fock space directly to local observables in real space through a probability-weighted all-pairs distance statistic.

Mira: And it gives us a tool to see exactly where the heterogeneity resides, whether it’s state level, sample level, or outer ensemble level.

Yu-Jing Liu, Chen Cheng

Lanzhou Center for Theoretical Physics · Key Laboratory of Quantum Theory and Applications of MoE · Key Laboratory of Theoretical Physics of Gansu Province · Gansu Provincial Research Center for Basic Disciplines of Quantum Physics

cond-mat.dis-nn, cond-mat.stat-mech

Submitted: 2026-09-14

Updated: 2026-11-03

Comments: 22 pages, 19 figures

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 90/100

The gist: Statistical levels and spatial modes of Fock-space heterogeneity in many-body localization crossovers is a statistical framework based on configuration-space distance distributions that uses a…

Key concepts

Configuration-Space Distance Distribution
This is a statistical tool that measures the separation between pairs of configurations drawn from the same eigenstate's probability distribution. It groups these distances to reveal how configurations are spread out in real space, allowing researchers to quantify spatial heterogeneity.
Variance Decomposition (VarP¯(RH))
This method breaks down the total variance of a measured quantity into three distinct parts: fluctuations within a single eigenstate (intra-eigenstate), variations between different eigenstates of one sample (state-to-state), and variations among different disorder realizations or samples (outer ensemble).
Outer Variance ($\chi_{outer}$)
This component specifically quantifies the variation that occurs between different disorder realizations or quasiperiodic phase samples. It is linked to the uniform spatial projection of local-memory fluctuations and is crucial for distinguishing between random and quasiperiodic systems.
Spatial Covariance ($C_{outer}(r)$)
This measures how the spatial organization of outer local-memory fluctuations changes with distance ($r$). The analysis shows that random outer fluctuations have a strong uniform component, while quasiperiodic phase fluctuations are more organized at specific wave numbers and tend to cancel out in the average.

Terminology

Summary

Statistical levels and spatial modes of Fock-space heterogeneity in many-body localization crossovers is a statistical framework based on configuration-space distance distributions that uses a variance decomposition to separate fluctuations within eigenstates, between eigenstates of one sample, and between samples. p. 0 At the exact-diagonalization sizes studied, the clearest difference occurs among disorder realizations or quasiperiodic phase samples, whereas within-eigenstate and within-sample eigenstate-to-eigenstate contributions remain broadly comparable. p. 0 Spatial covariances show that random outer fluctuations have a much stronger uniform component, whereas quasiperiodic phase fluctuations are organized more strongly at finite wave number and partly cancel in the spatial average controlling the distance center. p. 0

Pruning redundant reasoning tokens

The framework connects Fock-space probability structure to local observables in real space by using a probability-weighted all-pairs distance statistic that groups pairs of configurations drawn independently from the same eigenstate’s configuration probability distribution by their separation to form a configuration-space distance distribution. This construction allows for the separation of three sources of center and distance variation: intra-eigenstate variation, state-to-state variation within one disorder realization or quasiperiodic phase sample, and outer-ensemble variation among those samples. p. 1 The total variance is decomposed using the law of total variance successively to yield VarP¯(RH) ≡ χintra + χstate + χouter. p. 3

How the training signal is built

The analysis involves comparing random-cosine (RC) and quasiperiodic-cosine (QP) Ising ensembles with the same one-site field marginal to identify where their difference resides. The contrast emerges at the outer level, where drift among RC disorder realizations is enhanced over a broad part of the crossover, whereas drift among QP phase samples is smaller. p. 4 The outer variance of the distance center is determined by this uniform component, which relates to the uniform spatial projection of sample-to-sample local-memory fluctuations. p. 6

Where the gains come from

The analysis reveals that predominantly positive correlations between local-memory variations at different sites enhance distance-center fluctuations in RC, while in QP, a larger fraction of the variation takes the form of spatial modulation that cancels in the uniform projection. The realized field amplitude provides a simple physical explanation for the collective RC response, as near the crossover, the ensemble-mean distance center is highly sensitive to field strength. p. 0 The outer variance of the distance center is determined by this uniform component, which depends on both the magnitude of local variations and their cross-site correlations. p. 6

Spatial organization of outer local-memory fluctuations

The spatial organization is characterized by the covariance Couter(r) = 1/(L − r) sum i=1 to L X−r i=1 Γ outer i,i+r. The continuous-wave-number curve in Fig. p. 7 5(d) displays the spectral shape, where k = 0 is the strongest individual RC mode while the QP maximum remains near the nonzero wave number kh2. p. 7 This indicates that QP phase variation can produce appreciable local responses that partly cancel in the uniform spatial projection, yielding a smaller χouter. p. 6

Full-distribution distinguishability

The Jensen–Shannon divergences measure the information lost when a label is hidden rather than the dispersion of a scalar center. The state-level contribution Jstate quantifies how much information a sampled distance carries about the eigenstate label when the physical sample is known, and Jouter quantifies its information about the physical sample label. p. 4 These measures obey the exact chain-rule identity S[P¯] = 1/NoutNE Xa,n S[Pan] + Jstate + Jouter. p. 3 The RC–QP contrast is resolved both in the outer variance of the distribution centers and in the distinguishability of the complete profiles. p. 8

Nested heterogeneity in a fixed-magnetization XXZ chain

When restricted to a fixed-S z = 0 sector, the state and outer differences resolved under this restriction reflect heterogeneity within one magnetization sector. The nested hierarchy therefore resolves state-to-state and sample-to-sample heterogeneity within a symmetryconstrained many-body sector. p. 3 The QP outer curves in panel (f) have much smaller outer variances and exhibit repeated crossings and nonmonotonic size ordering.

Conclusion

Configuration-space statistical geometry provides a direct way to distinguish sources of fluctuations in manybody ensembles. The intraeigenstate and within-sample state-to-state channels remain broadly comparable between ensembles. p. 1 The fixedmagnetization XXZ application extends this statistical distinction to a constrained configuration space. p. 0 Future work could resolve configuration differences within separate real-space segments and examine how each segment and the correlations between segments contribute to distance fluctuations across the full chain. p. 1

Acknowledgments

This research was supported by the National Natural Science Foundation of China (Grant Nos. 12174167, 12474139, 12247101), the Fundamental Research Funds for the Central Universities (Grant No. p. 0 lzujbky-2025-jdzx07), and the Natural Science Foundation of Gansu Province (No. p. 0 25JRRA799).

Appendix A

The distribution in Eq. (1) is equivalent to the ordered all-pairs distance distribution of Roy and Logan. p. 2 The related construction of Orito and Imura does not distinguish the two pair orderings and treats the zero shell separately. p. 2 The second inverse participation ratio in this basis is Pψ(0) = X s pψ(s) squared = IPR squared. p. 2

Appendix B

The matched-cosine calculation uses the open Ising chain of Eq. (11), with clean bonds, J = g = 1, and L = 8,. p. 4.., 15. For each converged matched-Ising sample at L = 8,. p. 4.., 15, we retain the NE = 100 eigenstates closest to E = 0. p. 4 The exact trace center of this sector for the stated open chain with Jz = 1 is Etr = −0. p. 5 p. 525, so σ = 0 is a common numerical target rather than an exact trace-centered energy. p. 0 p. 0

Appendix C

The benchmark displays both nested identities in a second random model: a sizable intra-eigenstate background can coexist with enhanced state and outer channels through the crossover. The RC–QP comparison with matched one-site marginals and clean bonds in Sec. p. 4 III isolates the effect of spatial field organization. p. 8

Appendix D

The main comparisons use finite outer populations and fixed eigenstate windows. The RC outer channel remains larger than its QP counterpart at the selected representative points. p. 5 p. 5 The QP outer values retain visible window sensitivity. p. 0 p. 0 p. 0 p. 0 p. 0

Appendix E

The Walsh–Hadamard step reorganizes the complete probability-vector pair sum without stochastic configuration-pair sampling. The first two moments of the binned distribution must also reproduce Eqs. p. 2 p. 2 p. 2 p. 2 p. 2 p. 2 p. 2 p. 2 p. 2 p. 2 (2) and (3).

Appendix F

The matched cosine ensembles have the same singlesite field distribution but different multisite correlations. The bare kernel is organized at the field wave number. p. 0 The bare-field identities provide a reference for the spatialstructure comparison. p. 6 p. 6

Appendix G

The Hamming distance remains the mismatch count in the selected basis. Within a constrained sector it need not coincide with the shortest-path distance on the graph of nonzero off-diagonal Hamiltonian matrix elements.

Appendix H

The Walsh–Hadamard step encodes spin configurations si = ±1 as bit strings x ∈ 0, 1 L, with xi = (1 − si)/2. The ordered pair sum can then be reorganized as the XOR autocorrelation RH(x, y) = w(x ⊕ y). p. 2 p. 2 p. 2

Appendix I

The first two moments of the binned distribution must also reproduce Eqs. (2) and (3). The exact trace center of this sector for the stated open chain with Jz = 1 is Etr = −0.25, so σ = 0 is a common numerical target rather than an exact trace-centered energy.

Appendix J

The two Jensen–Shannon terms are weighted Jensen–Shannon divergences: they quantify the information lost upon suppressing, respectively, the state label and the outer label. The mean-response interpolation, held-out prediction, and nonlinear mapping in Fig. p. 9 9 are described in Appendix B 2. p. 2

Appendix K

The bare kernel is organized at the field wave number. The bare-field identities provide a reference for the spatialstructure comparison.

Appendix L

The Hamming distance remains the mismatch count in the selected basis. Within a constrained sector it need not coincide with the shortest-path distance on the graph of nonzero off-diagonal Hamiltonian matrix elements.

Appendix M

The Walsh–Hadamard step encodes spin configurations si = ±1 as bit strings x ∈ 0, 1 L, with xi = (1 − si)/2. The ordered pair sum can then be reorganized as the XOR autocorrelation RH(x, y) = w(x ⊕ y).

Appendix N

The first two moments of the binned distribution must also reproduce Eqs. (2) and (3). At sizes where direct enumeration is practical, the complete binned distribution is compared with the ordered-pair sum in Eq. p. 4 p. 4 (1).

Appendix O

The Walsh–Hadamard normalization and checks encode spin configurations si = ±1 as bit strings x ∈ 0, 1 L, with xi = (1 − si)/2. The first two moments of the binned distribution must also reproduce Eqs. (2) and (3).

Appendix P

The Hamming distance remains the mismatch count in the selected basis. Within a constrained sector it need not coincide with the shortest-path distance on the graph of nonzero off-diagonal Hamiltonian matrix elements.

Appendix Q

The Walsh–Hadamard step reorganizes the complete probability-vector pair sum without stochastic configuration-pair sampling. The first two moments of the binned distribution must also reproduce Eqs. (2) and (3).

Appendix R

The Hamming distance remains the mismatch count in the selected basis. Within a constrained sector it need not coincide with the shortest-path distance on the graph of nonzero off-diagonal Hamiltonian matrix elements.

Appendix S

The Walsh–Hadamard step encodes spin configurations si = ±1 as bit strings x ∈ 0, 1 L, with xi = (1 − si)/2. The ordered pair sum can then be reorganized as the XOR autocorrelation RH(x, y) = w(x ⊕ y).

Appendix T

The first two moments of the binned distribution must also reproduce Eqs. (2) and (3). At sizes where direct enumeration is practical, the complete binned distribution is compared with the ordered-pair sum in Eq. (1).

Appendix U

The Walsh–Hadamard normalization and checks encode spin configurations si = ±1 as bit strings x ∈ 0, 1 L, with xi = (1 − si)/2. The first two moments of the binned distribution must also reproduce Eqs. (2) and (3).

Appendix V

The Hamming distance remains the mismatch count in the selected basis. Within a constrained sector it need not coincide with the shortest-path distance on the graph of nonzero off-diagonal Hamiltonian matrix elements.

Appendix W

The Walsh–Hadamard step reorganizes the complete probability-vector pair sum without stochastic configuration-pair sampling. The first two moments of the binned distribution must also reproduce Eqs. (2) and (3).

Appendix X

The Walsh–Hadamard normalization and checks encode spin configurations si = ±1 as bit strings x ∈ 0, 1 L, with xi = (1 − si)/2. The first two moments of the binned distribution must also reproduce Eqs. (2) and (3).

Appendix Y

The Hamming distance remains the mismatch count in the selected basis. Within a constrained sector it need not coincide with the shortest-path distance on the graph of nonzero off-diagonal Hamiltonian matrix elements.

Appendix Z

The Walsh–Hadamard step reorganizes the complete probability-vector pair sum without stochastic configuration-pair sampling. The first two moments of the binned distribution must also reproduce Eqs. (2) and (3).

Appendix AA

The Walsh–Hadamard normalization and checks encode spin configurations si = ±1 as bit strings x ∈ 0, 1 L, with xi = (1 − si)/2. The first two moments of the binned distribution must also reproduce Eqs. (2) and (3).

Appendix BB

The Walsh–Hadamard step reorganizes the complete probability-vector pair sum without stochastic configuration-pair sampling. The first two moments of the binned distribution must also reproduce Eqs. (2) and (3).

Appendix CC

The Walsh–Hadamard normalization and checks encode spin configurations si = ±1 as bit strings x ∈ 0, 1 L, with xi = (1 − si)/2. The first two moments of the binned distribution must also reproduce Eqs. (2) and (3).

Appendix DD

The main comparisons use finite outer populations and fixed eigenstate windows. The RC outer channel remains larger than its QP counterpart at the selected representative points. The QP outer values retain visible window sensitivity.

Appendix EE

The state channels and the RC outer channel change weakly between NE = 75 and 100 in Fig. 16. The smaller QP outer values are more sensitive for several sizes. p. 0

Appendix FF

The RC outer channel remains larger than its QP counterpart at the selected representative points. The QP outer values retain visible window sensitivity.

Appendix GG

The RC outer channel remains larger than its QP counterpart at the selected representative points. The QP outer values retain visible window sensitivity.

Appendix HH

The RC outer channel remains larger than its QP counterpart at the selected representative points. The QP outer values retain visible window sensitivity.

Appendix II

The state-to-state and outer-ensemble channels do not reorganize at the same native disorder scale. The balance between state-level and realizationlevel heterogeneity therefore shifts through the crossover. p. 8 p. 8

Appendix III

The RC and QP state-to-state responses in panels (a) and (b) remain on broadly similar scales through much of the common displayed range. The outer responses in panels (c) and (d) separate much more strongly. p. 5

Appendix IV

The state–outer crossover chronology does not reorganize at the same native disorder scale. The balance between state-level and realizationlevel heterogeneity therefore shifts through the crossover.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved analysis of many-body localization crossovers by developing a statistical framework based on configuration-space distance distributions that uses a variance decomposition to separate fluctuations within eigenstates, between eigenstates of one sample, and between samples. This allows for separating intra-eigenstate, within-sample eigenstate-to-eigenstate, and between samples contributions to understand heterogeneity.

  2. Enhanced ability to distinguish the physical origin of ensemble contrast by employing nested statistics, specifically the Jensen–Shannon divergence measures Jstate and Jouter, which quantify the distinguishability of distance distributions at different statistical levels. This allows researchers to determine where in this sampling hierarchy the matched random–quasiperiodic difference becomes pronounced.

  3. Resolution of spatial organization within sample-to-sample fluctuations by analyzing spatial covariances, showing that "random outer fluctuations have a much stronger uniform component, whereas quasiperiodic phase fluctuations are organized more strongly at finite wave number and partly cancel in the spatial average controlling the distance center. This connects Fock-space probability geometry to real space structure of sample-dependent local memory."

  4. Prediction of sample-to-sample distance center fluctuations using an amplitude-response analysis, where the outer variance is determined by a uniform projection of the localmemory covariance, which depends on both the magnitude of local variations and their cross-site correlations. This provides a physical explanation for why random realizations sample different overall field strengths leading to enhanced sample-to-sample fluctuations.

  5. Characterization of spatial modes in outer fluctuations by calculating the Fourier spectrum Souter(k) and comparing it to the bare field wave number kh2, showing that the QP maximum remains near the nonzero wave number kh2, which is related to the interacting squared-local-polarization response. This helps identify how sample-to-sample fluctuations of halfchain entanglement likewise reflect the combined effects of field-amplitude fluctuations and its own mean response.

Related papers