Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

For a spectrum that is a statistical mixture of symmetry sectors, the largest correlated subsequence ('characteristic symmetry sector') has size equal to the size-biased average of the sector dimensions; removing Poisson gaps recovers it.

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 ↗

For a spectrum that is a statistical mixture of independent symmetry sectors, the correlated remainder after removing Poisson-distributed gaps has expected size equal to the size-biased average of the sector dimensions.

T0 review reviewed 2026-08-02 challenge →

load-bearing objection Eq. (7) is a clean repackaging of known mixture statistics, but the paper never shows the decimation output actually equals that formula — the PF test is a good check, the abstract oversells, and the MBL analysis quietly drops its own assumptions. the 4 major comments →

arxiv 2602.20256 v2 pith:SDUHGYSL submitted 2026-02-23 cond-mat.stat-mech cond-mat.dis-nnquant-ph

Spectral Decimation of Quantum Many-Body Hamiltonians

Feng He , Arthur Hutsalyuk , Giuseppe Mussardo , Andrea Stampiggi This is my paper
classification cond-mat.stat-mech cond-mat.dis-nnquant-ph
keywords spectral decimationstatistical mixturecharacteristic symmetry sectorsize-biased averageHilbert-space fragmentationmany-body localizationlevel statisticscharacteristic symmetry entropy
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.

The reading

The paper seeks to establish spectral decimation as a quantitative, assumption-light probe of emergent symmetries in quantum many-body spectra. For a spectrum that is a statistical mixture of independent symmetry sectors, it derives a closed-form expression for the size of the characteristic symmetry sector (CSS)—the largest subsequence of levels with non-Poissonian correlations—namely the size-biased average of the sector dimensions, dout = Σ_i d_i²/d. This matters because many-body spectra routinely look Poissonian even when they are superpositions of chaotic blocks (Hilbert-space fragmentation) or of disorder-dependent sectors (many-body localization); the CSS recovers the statistically dominant correlated block underneath. The paper verifies the formula in the pair-flip fragmented model and uses it in the disordered Heisenberg chain, where a shrinking CSS with peaky strong-disorder gap statistics indicates that emergent integrability is effectively free, not generic integrable. It also defines a characteristic symmetry entropy that exhibits finite-size scaling collapse, supplying an inexpensive diagnostic of hidden structure.

Core claim

The central claim is that the output of the spectral decimation algorithm—iteratively removing gaps that are statistically indistinguishable from a Poisson process—is, for a statistical mixture of N disjoint sectors of dimensions d_i, a Gaussian random variable with expectation E[dout] = d(1−M) = Σ_i d_i²/d, where d = Σ_i d_i is the total dimension. This identity is derived from a small-gap expansion of the exact mixture gap statistics: near zero spacing the mixture density splits into a Poisson component M e^{−s}, coming from pairs of levels belonging to different sectors, and a correlated component that is the density-weighted superposition of the sectors' own repulsive distributions. Deci

What carries the argument

The central object is the characteristic symmetry sector (CSS): the largest subsequence of levels that exhibits non-Poissonian correlations (level repulsion or isolated peaks). The argument rests on two coupled pieces: (i) a small-gap approximation to the exact gap statistics of a statistical mixture, which splits the near-zero spacing density into a Poisson fraction M = 1 − Σ_i (ρ_i/ρ)² and a correlated remainder; and (ii) the spectral decimation algorithm, which uses rejection sampling with a target Poisson density to remove Poisson-compatible gaps at each iteration, controlled by an extraction fraction f and a halting threshold d_halt. The load-bearing identity is Eq. (7), E[dout] = Σ_i d

Load-bearing premise

Every symmetry sector must exhibit level repulsion on its own, and the relative densities of sectors must stay fixed while the decimation thins the sample; if a sector has Poisson spacings or the mixture weights drift, the size-biased formula ceases to control the outcome.

What would settle it

Construct a mixture of one GOE block of dimension d_G and one independent Poisson-distributed block of dimension d_P, so the size-biased prediction is E[dout] = (d_G² + d_P²)/d. If the mean dout over many realizations departs from this value in a way that grows with d_P—or if the remainder gap statistics lose their level repulsion as d_P is increased—then the assumption that Poisson gaps come only from pairs of levels in different sectors is wrong. This is directly testable with the released decimation code.

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

If this is right

  • For any spectrum that is a statistical mixture, the decimated remainder's size gives a direct, symmetry-blind estimate of Σ_i d_i²/d, so the effective sector structure can be read off from a single pooled sample.
  • In fragmented systems, chaotic subsectors can be isolated even when the global spectrum is almost entirely Poisson; the pair-flip model matches the theoretical dout ≈ 2^L/√(πL).
  • In the disordered Heisenberg chain, the CSS shrinks exponentially with disorder strength and system size, and its strong-disorder gap statistics show the peaked form of effectively free spectra—so MBL's emergent integrability is a mixture phenomenon, not a set of independent levels.
  • The characteristic symmetry entropy collapses onto a single curve under the scaling ℓ = L^{1/ν}(W−W*) with W* ≈ 3.3 and ν ≈ 0.8 (at ΔE=0.9), giving a computable, system-size-dependent estimate of the MBL crossover location and exponent, with the caveat that these are crossover rather than thermodynamic quantities.
  • The gap-only decimation distinguishes integrable, mixed, and free spectra without any dynamical or entanglement data, because the three leave qualitatively different remainder statistics: Poisson, repulsive, and peaked.

Where Pith is reading between the lines

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

  • Because dout = Σ d_i²/d is a participation ratio of the sector-dimension distribution, an immediate unstated consequence is that one can define an effective number of sectors N_eff = d²/Σ d_i²; a single decimation run then yields N_eff directly, which could be compared with symmetry data from the Hamiltonian's commutant algebra.
  • The quantitative disagreement between gap-based and r-ratio-based CSS sizes suggests that three-level correlations obey a different mixture law; deriving it would extend the method to observables that need no spectral unfolding and might sharpen the MBL crossover estimate.
  • The sub-Harris exponent and the paper's own Kosterlitz–Thouless caveat raise the possibility that the CSE crossover is governed by rare regions or finite-size effects; applying the same method to quasiperiodic potentials—where the MBL transition is believed to be sharper—would test whether the scaling form is universal or disorder-dependent.
  • A decisive, cheap test of the level-repulsion assumption is to run the decimation on a constructed mixture of one GOE block and one small Poisson block of known size: if the measured dout systematically exceeds or falls short of the size-biased value as the Poisson block's weight grows, the assumption that Poisson gaps arise only between sectors would be falsified.
Share X Bluesky LinkedIn Reddit HN

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

4 major / 6 minor

Summary. The paper develops an analytical description of spectral mixtures and a spectral decimation algorithm, claiming that the size of the extracted 'characteristic symmetry sector' (CSS) is given by the size-biased average of the sector dimensions, Eq. (7). The authors derive a small-gap approximation for the mixture gap PDF, Eq. (6), and propose the CSS as the correlated remainder after iteratively removing Poisson-like gaps. They apply the method to two models: a pair-flip model with Hilbert-space fragmentation, where Eq. (7) is checked against known sector dimensions, and the disordered Heisenberg chain, where the CSS shrinks with disorder and a 'characteristic symmetry entropy' (CSE) is introduced as a finite-size scaling observable. The main claims are: (i) spectral decimation provides an unbiased estimator of the CSS size; (ii) the CSS diagnoses emergent symmetries in fragmented and MBL spectra; (iii) the CSE shows a finite-size scaling collapse with effective parameters W* ~ 3.3–3.8 and nu < 1.

Significance. If the central claim were rigorously established, this would be a valuable addition to spectral diagnostics for quantum many-body systems. The paper has genuine strengths: the small-gap mixture approximation is derived from standard random-matrix results without fitting the central prediction; the pair-flip model provides a nontrivial check of Eq. (7) against independently known sector dimensions; the algorithm is publicly available and the code availability is stated; and the authors are careful in Section 4.3 to present the FSS as phenomenological and to flag the Harris-bound violation. The proposed CSE is a new observable that could be useful for detecting emergent integrability. However, the central step connecting the decimation output to Eq. (7) is not rigorously controlled, and the paper's own text acknowledges limitations that bear on this connection. The significance of the paper is therefore conditional on closing this gap.

major comments (4)
  1. [Section 2, Eq. (7) and Section 3, step 4] Eq. (7) states E[d_out] = d(1-M) = sum_i d_i^2/d, but the decimation output d_out is defined by an iterative protocol that removes a fixed fraction f of the current remainder based on the empirical zero-bin value, halting when π_0 < f or d_R <= d_halt. The analytical derivation of Eq. (6) only gives the small-s form of the mixture PDF; it does not show that the stochastic algorithm's stopping point equals d(1-M). Indeed, Section 3.1 H1 implies that for uncorrelated spectra d_out = d_halt, whereas the text states E[d_out] = 1 for such spectra (Section 2, paragraph before Eq. (7)). These are consistent only if d_halt = 1, but the paper explicitly uses d_halt = n in pooled samples. Thus Eq. (7) is not shown to govern the algorithm output in general; the agreement with the PF model and Appendix C could be partly by construction, since the estimated Poisson fraction and the mixture fraction M
  2. [Section 2, assumption p_i(s) ~ s^{nu_i}] Eq. (6) and the identification of the Poisson component rely on the assumption that each sector has level repulsion with nu_i > 0. The paper itself identifies this limitation in Section 2 and applies the framework to the MBL regime, where at large disorder the remainder statistics show clear peaks and p_i(0) > 0 (Section 4.2.1 and Fig. 3(c)), i.e., the very assumption is violated. In that regime the size-biased formula for d_out is uncontrolled. The text does not provide a separate derivation for spectra with peaked gap distributions, so the claim that the CSS remains a valid estimator of the size-biased average in the strong-disorder regime is unsupported."
  3. [Definitions in Abstract and Sections 1 and 5] The abstract defines the CSS as the 'largest subsequence of levels exhibiting non-Poissonian correlations', but Eq. (7) computes a size-biased average of sector dimensions, not the dimension of a largest subsequence. The two notions coincide only for sectors of equal dimension. The paper uses 'largest' loosely and the discrepancy is not discussed. This issue does not affect the derivation of Eq. (7) but it makes the abstract's definition misleading and could be fixed by rephrasing to 'characteristic' or 'dominant' sector."
  4. [Section 4.3.1 and Eq. (21)] The FSS analysis uses the phenomenological ansatz Eq. (21) with parameters a,b,c,W*,nu. The paper explicitly states (Section 4.3) that this FSS is not based on a renormalization-group form and that the system sizes may not be in the thermodynamic regime. Nevertheless, the extracted nu < 1 violates the Harris bound, and the text acknowledges that the KT form does not collapse the data. These caveats are appropriate, but the result is presented as a 'universal finite-size scaling collapse' in the conclusions. That overstatement should be softened, and the phenomenological character of the fit should be stated in the abstract or conclusions."
minor comments (6)
  1. [Section 2, Eq. (2)] The notation d = sum_i d is used without defining the subscript i on d_i in Eq. (2); it would be clearer to write d = sum_i d_i.
  2. [Appendix B, Eq. (48) referenced as Eq. (43)] In Appendix C, the text says 'variance predicted from the spectral decimation, Eq. (43)' but the variance formula is Eq. (47) or (48). The citation numbering is inconsistent.
  3. [Section 3, step 2] The acceptance ratio M(m) is defined as min{π_0, 1} where π_0 is the empirical zero-bin PDF, but the relationship between this quantity and the exact Poisson fraction M in Eq. (6) is not made explicit beyond 'up to order O(delta)'. This is a clarity issue that affects reproducibility.
  4. [Figure 3 caption] Figure 3 lists 'L = 16, W = 6' in the caption but the text refers to different W values; please verify consistency.
  5. [Section 4.1, Eq. (16)] The binomial identity sum_M D^2_{L,M} = binom(2L,L) is stated without proof; a brief derivation or reference would help the reader.
  6. [Section 4.2, parameter range] The text says 'L = 16, W = 6' in a figure caption while the main text says 'L = 9,...,16' and 'W = 6.5' in other places; please standardize the disorder values.

Circularity Check

3 steps flagged

Eq. (7) partly reduces to the paper's own definition of the CSS, and the decimation output is tied to the same zero-bin statistic and to f/d_halt, making the central identification partially by construction.

specific steps
  1. self definitional [Section 2, after Eq. (6) and Eq. (7)]
    "Generally, this Poisson component of the mixture is a fraction M of it, which also corresponds to the value of the PDF near the origin. What remains, of size(1−M) is the correlated component of the mixture, the 'characteristic symmetry sector' (CSS). The spectral decimation, applied to a mixture, separates the Poisson component from the CSS. Its output, a spectrum of size dout, will then be distributed according to the correlated component of Eq. (6). The size of the CSS extracted through the spectral decimation is a Gaussian random variable with an expectation value E[dout] = d(1−M) = Σ_i d_i"

    The CSS is here defined as the leftover fraction 1−M of the mixture formula, and then the central 'prediction' E[dout]=d(1−M) is simply that definition re-stated as the decimation output. No equation in Section 3.1 connects the algorithm's halting rules (which use f, d_halt and the empirical zero-bin) to the algebraic sector dimensions Σ d_i^2/d; the equality is imposed rather than derived from the decimation steps.

  2. fitted input called prediction [Section 3.1, algorithm step 2 and halting condition H2]
    "To the set R_m, one applies rejection-sampling ... with target probability q˜(s)=e^{−s} and acceptance ratio M(m)=min{π0,1}. Here M(m) is the fraction of Poisson gaps, up to order O(δ), at iteration m, estimated from the zero-bin empirical PDF π0. ... If after 3. and d_R(m)>d_halt, the number of accepted gaps, whose expected value is E[d_A(m)]=M(m)d_R(m), is less than d_E(m)=f d_R(m), the algorithm halts."

    The theoretical M in Eq. (6) is identified as 'the value of the PDF near the origin' (the zero-bin), and the algorithm estimates M from the same zero-bin π0 and halts when this estimate falls below f. Thus the output dout is calibrated to the same zero-bin quantity that appears in d(1−M), while also depending explicitly on f and d_halt (H1 gives dout=d_halt for Poisson spectra unless d_halt=1). The agreement of dout with Eq. (7) is therefore partly built into the stopping rule rather than being an independent consequence of the decimation.

  3. other [Appendix B.5, 'Variance of the Output after Resamplings']
    "Let the expectation value of the output be dout = D, implying that the Poisson fraction is M = 1 − D/d."

    The variance calculation assumes the very identification it is meant to support: D=d(1−M) is used to define M from D, so the mean relation E[dout]=d(1−M) is taken as input rather than derived from the iterative removal process. This makes the later Gaussian/variance statements conditional on the asserted equality rather than independent evidence for it.

full rationale

The small-gap mixture approximation, Eq. (6), is a legitimate derivation from standard Porter–Rosenzweig/Berry–Robnik/Mehta formulas, and the algebraic step to Eq. (7), d(1−M)=Σ d_i^2/d, is correct given the definition M=1−Σ(ρ_i/ρ)^2. The circularity lies in the unexplained identification of the spectral-decimation output with this quantity. In Section 2 the CSS is defined as the d(1−M) 'correlated component', and then the paper asserts that the decimation output has expectation d(1−M); but the algorithm's own halting conditions (H1/H2) depend on the free parameters f and d_halt and on the empirical zero-bin π0, not on the sector dimensions. Indeed, H1 would give dout=d_halt for a perfectly Poisson spectrum, so E[dout]=1 requires a special choice d_halt=1, exposing that Eq. (7) is not a direct consequence of the algorithm. Appendix B.5 similarly assumes D=d(1−M) when it writes M=1−D/d. The numerical checks in Appendix C and the pair-flip model are genuine external benchmarks with known sector sizes, and they give the identification empirical support; this prevents the paper from being fully circular. The result nevertheless has a substantial component that is definitional and algorithm-calibration-dependent, so the central claim is only partially independent of its own construction.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 2 invented entities

The central formula uses standard mixture theory plus the assumption that each sector repels levels. The algorithm-to-formula equivalence is an unproved assumption. The MBL/CSE scaling introduces a phenomenological tanh fit. No microscopic physical entities are introduced; CSS and CSE are operational statistics, not physical degrees of freedom.

free parameters (5)
  • extraction fraction f = f = π0,GOE(δ) + 1.96 σGOE,0 (single samples); f = π0,GOE(δ) (pooled samples)
    Algorithm parameter controlling how many Poisson gaps are removed per iteration; chosen by hand from GOE zero-bin behavior (Appendix B.3). Changing f changes d_out, and the paper argues only that suitable choices avoid systematic bias.
  • halting threshold d_halt = d_halt = n for pooled samples; otherwise α = 1 − d_halt/d
    Stops the decimation and sets the confidence level. The paper claims robustness to moderate variations but does not quantify full sensitivity.
  • FSS fit parameters a,b,c,W*,ν = a=0.0528(33), b=0.3093(47), c=0.00955(70), W*=3.321(52), ν=0.824(22) at ∆E=0.90
    Fit constants in the tanh collapse Eq. (21); not derived from a Hamiltonian or RG. The authors note ν violates the Harris bound and treat the values as effective crossover quantities.
  • resampling count r = 25
    Number of decimation repeats used to average d_out; affects statistical errors, and no convergence study is shown.
  • energy window ΔE = 0.3, 0.6, 0.9
    Analysis choice that changes W* and ν (Fig. 5); the paper uses it to compensate for limited system sizes without a first-principles prescription.
axioms (5)
  • domain assumption Each sector p_i(s) has Brody-like repulsion: p_i(s)∼s^{ν_i}, ν_i>0, so Poisson gaps arise only from pairs of levels in different sectors.
    Used to derive the small-s approximation Eq. (6); fails for sectors that are themselves Poisson (ν=0), such as regular Berry-Robnik components.
  • standard math The exact mixture gap PDF is the product formula Eq. (2) from Porter–Rosenzweig, Berry–Robnik, and Mehta.
    The paper cites these results and builds the entire derivation on them without proving them.
  • ad hoc to paper Rejection sampling with target e^{-s} extracts exactly the Poisson component, and the remainder is the correlated CSS with size d(1−M).
    This is the bridge between Eq. (6) and Eq. (7); the paper gives numerical checks but no theorem that the algorithm output obeys this law.
  • ad hoc to paper The CSE obeys a tanh finite-size scaling form Eq. (21).
    Phenomenological; the authors state it is not RG-derived and that a KT form does not collapse the data.
  • domain assumption In the disordered XXZ chain, emergent LIOMs are the origin of the shrinking CSS.
    Interpretation relies on the MBL/LIOM literature; footnote 2 acknowledges the finite-size-regime caveat.
invented entities (2)
  • Characteristic symmetry sector (CSS) no independent evidence
    purpose: Statistical construct representing the correlated remainder after removing Poisson-distributed gaps; used to infer the effective symmetry sector size.
    No falsifiable prediction outside the decimation procedure; the claimed link to physical symmetry sectors is the paper's hypothesis.
  • Characteristic symmetry entropy (CSE) no independent evidence
    purpose: Finite-size scaling observable Σ(W,L) = (1/L) log_2(log d − log d_out), proposed as an order parameter for emergent integrability.
    It is a derived summary of the CSS; its universal collapse is demonstrated only through a fitted ansatz.

reviewed 2026-08-02 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Spectral Decimation of Quantum Many-Body Hamiltonians." pith.science (2026). https://pith.science/paper/SDUHGYSL

@misc{pith2026260220256,
  author       = {Pith},
  title        = {Pith review of: Spectral Decimation of Quantum Many-Body Hamiltonians},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SDUHGYSL}},
  note         = {Machine review of arXiv:2602.20256}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We develop a systematic theory of spectral decimation for quantum many-body Hamiltonians and show that it provides a quantitative probe of emergent symmetries in statistically mixed spectra. Building on an analytical description of statistical mixtures, we derive an explicit expression for the size of a characteristic symmetry sector (CSS), defined as the largest subsequence of levels exhibiting non-Poissonian correlations. The CSS dimension is shown to be the size-biased average of the underlying symmetry sectors, establishing a direct link between spectral statistics and Hilbert-space structure. We apply this framework to two paradigmatic settings: Hilbert-space fragmentation and disorder-induced many-body localization (MBL). In fragmented systems, the CSS reproduces the mixture prediction and isolates correlated subsectors even when the full spectrum appears nearly Poissonian. In the disordered Heisenberg chain, spectral decimation reveals the gradual emergence of integrability through a shrinking CSS, whose statistics exhibit signatures consistent with local integrals of motion. We introduce a characteristic symmetry entropy (CSE) as a finite-size scaling observable and extract, within accessible system sizes, the crossover exponents. Our results establish spectral decimation as a controlled, unbiased and computationally inexpensive diagnostic of hidden structure in many-body spectra, capable of distinguishing between chaotic dynamics, statistical mixtures, and emergent integrability.

Figures

Figures reproduced from arXiv: 2602.20256 by Andrea Stampiggi, Arthur Hutsalyuk, Feng He, Giuseppe Mussardo.

Figure 1
Figure 1. Figure 1: Outcome of the decimation in the case the parent Hamiltonian is a) integrable, b) statistical mixture [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Application of spectral decimation to the pair-flip model of Eq. ( [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Histograms of gaps and r-ratios of the one-dimensional disordered Heisenberg chain for L = 16 at weak (W = 0.5) and strong (W = 6.5) disorders, in a energy window of ∆E = 90% of the full spectrum. At weak disorder, one observe excellent agreement of the distribution of gaps (a) and r-ratios (b) with the Wigner-Dyson distribution of GOE random matrices. The value of dout/d is approximately 1 in both cases, … view at source ↗
Figure 4
Figure 4. Figure 4: (a) Plot of the CSE over a large window of the spectrum ( [PITH_FULL_IMAGE:figures/full_fig_p015_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Results of finite-size scaling (FSS) of the characteristic symmetry entropy (CSE). (a) the FSS of [PITH_FULL_IMAGE:figures/full_fig_p017_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Statistical Mixture of N GOEs of the same size dGOE = 104 , respectively N = 2 (a), N = 4 (b) and N = 8 (c). At moderately small s ≲ 0.5, the small s approximation, see Section 2, qualitatively describes the gap distribution, whose exact PDF [22, 60, 62] is shown. In particular, it predicts the correct value of the PDF near the origin. As N increasing, the Poisson component of the mixture becomes dominant … view at source ↗
Figure 7
Figure 7. Figure 7: Decimation of a single realization of a statistical mixture, with [PITH_FULL_IMAGE:figures/full_fig_p024_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Result of the decimation procedure on statistical mixtures generated by [PITH_FULL_IMAGE:figures/full_fig_p025_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.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Level statistics of the disordered Haldane-Shastry model with $1/r^\alpha$ interaction

    cond-mat.str-el 2026-04 unverdicted novelty 5.0

    In the long-range Haldane-Shastry model, pristine Poisson level statistics emerge only with combined position disorder and random magnetic fields, with an approximate scaling collapse governed by the product αδ when S...

Reference graph

Works this paper leans on

6 extracted references · 1 canonical work pages · cited by 1 Pith paper · 1 internal anchor

  1. [1]

    1S. R. White,Density matrix formulation for quantum renormalization groups, Phys. Rev. Lett.69, 2863–2866 (1992). 2S. R. White,Density-matrix algorithms for quantum renormalization groups, Phys. Rev. B 48, 10345–10356 (1993). 3R. Gezzi, T. Pruschke, and V. Meden,Functional renormalization group for nonequilibrium quantum many-body problems, Phys. Rev. B75...

    Pith/arXiv arXiv 1992

  2. [2]

    Spectroscopy, Phys. Rev. X11, 031034 (2021). 14D.-L. Deng, X. Li, and S. Das Sarma,Quantum Entanglement in Neural Network States, Phys. Rev. X7, 021021 (2017). 15K. Hornik,Approximation capabilities of multilayer feedforward networks, Neural Networks 4, 251–257 (1991). 16G. Carleo and M. Troyer,Solving the quantum many-body problem with artificial neural ...

    2021

  3. [3]

    Repulsion of Energy Levels

    Chiribella, D. Tao, et al.,Artificial intelligence for representing and characterizing quantum systems, arXiv preprint arXiv:2509.04923 (2025). 19R. Rende, L. L. Viteritti, F. Becca, A. Scardicchio, A. Laio, and G. Carleo,Founda- tion neural-networks quantum states as a unified Ansatz for multiple hamiltonians, Nature Comm.16(2025). 20F. He, A. Hutsalyuk,...

    Pith/arXiv arXiv 2025

  4. [4]

    Coexist, J. Phys. A: Math. Gen.17, 2413–2421 (1984). 63G. De Tomasi, D. Hetterich, P. Sala, and F. Pollmann,Dynamics of strongly interacting sys- tems: From Fock-space fragmentation to many-body localization, Phys. Rev. B100, 214313 (2019). 64T. Rakovszky, P. Sala, R. Verresen, M. Knap, and F. Pollmann,Statistical localization: From strong fragmentation t...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.cond-mat/0602510 1984

  5. [5]

    Bloch, and C

    Huse, I. Bloch, and C. Gross,Exploring the many-body localization transition in two dimen- sions, Science352, 1547–1552 (2016). 103R. Vosk, D. A. Huse, and E. Altman,Theory of the many-body localization transition in one-dimensional systems, Phys. Rev. X5, 031032 (2015). 104K. Agarwal, E. Altman, E. Demler, S. Gopalakrishnan, D. A. Huse, and M. Knap,Rare-...

    Pith/arXiv arXiv 2016

  6. [6]

    Chains, Phys. Rev. X13, 011041 (2023). 136M. Žnidarič, T. Prosen, and P. Prelovšek,Many-body localization in the HeisenbergXXZ magnet in a random field, Phys. Rev. B77, 064426 (2008). 137D. J. Luitz, N. Laflorencie, and F. Alet,Many-body localization edge in the random-field Heisenberg chain, Phys. Rev. B91, 081103 (2015). 138Á. L. Corps, R. A. Molina, an...

    Pith/arXiv arXiv 2023

This paper was first reviewed by deepseek-v4-flash on August 2, 2026.