The DOI matches its Crossref record; 49 of the 49 references it cites were found in a registry or an index; 5 of the 5 sources its claims cite were checked for support.
Local law and Tracy–Widom limit for sparse random matrices
Open accessOverview
CoChat Summary
TL;DR
The paper proves a local law up to the spectral edges for sparse random matrices and shows GOE Tracy–Widom fluctuations for the rescaled extremal eigenvalues once a deterministic sparsity-induced edge shift is included. For Erdős–Rényi adjacency matrices this gives Tracy–Widom fluctuations of the second largest eigenvalue for \(p \gg N^{-2/3}\) with a deterministic shift of order \((Np)^{-1}\).
Key findings
- Under Assumption 2.3 with \(N^\phi \le q \le N^{1/2}\), the normalized Green-function trace satisfies \(|m_H(z)-\tilde m(z)| \prec 1/q^2 + 1/(N\eta)\) uniformly on \(E\), where \(\tilde m\) solves \(1+z\tilde m+\tilde m^2+\frac{s^{(4)}}{q^2}\tilde m^4=0\) and is the Stieltjes transform of a deterministic symmetric measure \(\tilde\rho\) with support \([-L,L]\). The corresponding density estimate is \(|n(E_1,E_2)-n_{\tilde\rho}(E_1,E_2)| \prec \frac{E_2-E_1}{q^2}+\frac1N\).
- The edge location satisfies \(L=2+\frac{s^{(4)}}{q^2}+O(q^{-4})\), and the operator norm obeys \(|\,\|H\|-L\,| \prec 1/q^4+1/N^{2/3}\).
- If \(\phi>1/6\), i.e. \(q\gg N^{1/6}\), then \(\lim_{N\to\infty}\mathbb P(N^{2/3}(\lambda_1^H-L)\le s)=F_1(s)\), the GOE Tracy–Widom law. For the Erdős–Rényi adjacency matrix with \(p\gg N^{-2/3}\), the second largest eigenvalue satisfies \(\lim_{N\to\infty}\mathbb P(N^{2/3}(\lambda_2^A-L-a)\le s)=F_1(s)\), with \(L=2+d^{-1}+O(d^{-2})\), \(d=pN\), and shift \(a=\frac{q}{N}(1-q^2/N)^{-1/2}\).
- The Tracy–Widom result extends to joint distributions of the first \(k\) rescaled extreme eigenvalues, matching the GOE.
Methods
- The model is a real symmetric \(N\times N\) sparse random matrix with zero diagonal, independent off-diagonal entries, moment conditions \(\mathbb E H_{ij}=0\), \(\mathbb E H_{ij}^2=(1-\delta_{ij})/N\), and \(\mathbb E|H_{ij}|^k\le (Ck)^{ck}/(Nq^{k-2})\) for \(k\ge3\), with sparsity parameter \(N^\phi\le q\le N^{1/2}\). This includes centered Erdős–Rényi adjacency matrices and diluted Wigner matrices.
- The local law is proved via a new recursive moment estimate for the normalized trace \(m\) of the Green function, controlling high moments recursively using cumulant expansions rather than fully expanding all powers.
- The edge universality proof uses the Green function comparison method with a continuous flow interpolating between the sparse random matrix and the GOE, estimating the rate of change of \(m\) along the flow and offsetting it by the edge shift.
Limitations
- The paper restricts the sparsity parameter to \(N^\phi\le q\le N^{1/2}\) for arbitrarily small \(\phi>0\), though it expects the method can cover \((\log N)^{A_0\log\log N}\le q\le N^{1/2}\), \(A_0\ge30\), as considered in prior work.
- Edge universality requires \(\phi>1/6\); the paper expects the exponent \(1/6\) to be critical. For \(q\ll N^{1/6}\), bulk eigenvalue fluctuations are larger than \(N^{-2/3}\) and are expected to affect edge fluctuations, so the Tracy–Widom limit is not expected to hold. For Erdős–Rényi graphs, if \(p\ll N^{-2/3}\), an edge-universality-based test may fail.
- The assumption of zero diagonal entries is stated to be easily dropped. Evident from the provided text: the excerpt is truncated after Section 3, so the detailed proofs in Sections 4–7 are not available here for verification.
Model-written summary of a shortened version of the paper’s full text (it is longer than we read). Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 1 Not valid · 0 Not checked · 7 Own results and interpretations
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Proposition 2.8 (Lemma 4.4 of [10]). Suppose that H satisfies Assumption 2.3 with φ > 0. Then, |‖H‖ − 2| ≺ 1 q2 + 1 N2/3 . (2.25)”
What the sources say
- Could not verifycites [10] Che, Z.: Universality of random matrices with correlated entries. Electron. J. Probab. 22, 1–38 (2017) sourceExperimental — accuracy not yet measured
In §4, the numbered result is Theorem 4.4, a fixed-point theorem, not a lemma giving the claimed norm bound. The local law in §2.3 concerns the bulk of the spectrum; searches for Assumption 2.3 and the stated q⁻² + N⁻²ᐟ³ bound found no passage in our copy establishing this claim. Page numbers were unavailable. (LLM, conf=0.77)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Proposition 2.7 (Theorem 2.16 and Remark 2.18 in [10]). Suppose that H satisfies Assumption 2.3 with φ > 0. Denote by (uH i ) the ℓ2-normalized eigenvectors of H. Then, max 1≤i≤N ‖uH i ‖∞ ≺ 1√ N . (2.24)”
What the sources say
- Partly supportedcites [10] Che, Z.: Universality of random matrices with correlated entries. Electron. J. Probab. 22, 1–38 (2017) sourceExperimental — accuracy not yet measured
Corollary 2.12 (page unavailable) gives an eigenvector bound of order N⁻¹ᐟ² up to arbitrarily small powers of N, but only for indices whose classical locations satisfy ρ(γₖ) ≥ ω. The cited claim extends the bound to all eigenvectors; the paper’s discussion of Theorem 2.11 likewise describes delocalization for bulk eigenvalues, not every eigenvalue. (LLM, conf=0.89)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“max 1≤i,j≤N |GH ij (z)− δijmsc(z)| ≺ 1 q + √ Immsc(z) Nη + 1 Nη , (2.23)”
What the sources say
- Partly supportedcites [10] Che, Z.: Universality of random matrices with correlated entries. Electron. J. Probab. 22, 1–38 (2017) sourceExperimental — accuracy not yet measured
The local law in §2.3 (Theorem 2.11; page unavailable) gives a uniform entrywise resolvent estimate, but it compares G with a deterministic matrix M, not δᵢⱼm_sc, and uses a different error bound. The paper also notes in its discussion of Theorem 5.2 (page unavailable) that off-diagonal entries of G may be large. Our copy does not establish the stated bound under Assumption 2.3 with φ > 0. (LLM, conf=0.87)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“The main obstacle in the proof of the edge universality is that the local law obtained in [10] deteriorates at the edge of the spectrum.”
What the sources say
- Partly supportedcites [10] Che, Z.: Universality of random matrices with correlated entries. Electron. J. Probab. 22, 1–38 (2017) sourceExperimental — accuracy not yet measured
Theorem 2.11 (2.3 Main results; page unavailable) states the local law only for domains where the limiting density is bounded below, so it does not establish an edge local law. The paper does not say that its local law deteriorates at the edge or identify that as the main obstacle to proving edge universality. (LLM, conf=0.78)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Proposition 2.6 (Local semicircle law, Theorem 2.8 of [10]). Suppose that H satisfies Assumption 2.3 with φ > 0. Then, the following estimates hold uniformly for z ∈ E: |mH(z)−msc(z)| ≺ min { 1 q2 √ κ+ η , 1 q } + 1 Nη , (2.22)”
What the sources say
- Partly supportedcites [10] Che, Z.: Universality of random matrices with correlated entries. Electron. J. Probab. 22, 1–38 (2017) sourceExperimental — accuracy not yet measured
Section 2.3 (page unavailable) states a local law for the normalized trace of the resolvent, so the broad attribution is supported. But that result is Theorem 2.11, not Theorem 2.8, and it compares the trace with a model-dependent tr m under a positive-definite correlation assumption. It does not give the claimed semicircle comparison or the stated q- and κ-dependent bound. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
7 more claims
Create a free account to see all 12 claims and every source verdict.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 1. Introduction
“such that the expected number of non-vanishing entries is q2. For example for the adjacency matrices of the Erdős–Rényi graph we have q2 ≃ Np”
§ 2. Definitions and main results
“Note that the matrix A typically has N(N−1)p non-vanishing entries.”
Citations
Cites
Erdős, P., & Rényi, A. (2022) · Publicationes Mathematicae Debrecen · doi:10.5486/pmd.1959.6.3-4.12 · Cited by 5,337
Level-spacing distributions and the Airy kernel
Tracy, C. A., & Widom, H. (1994) · Communications in Mathematical Physics · doi:10.1007/bf02100489 · Cited by 1,828
Gilbert, E. N. (1959) · The Annals of Mathematical Statistics · doi:10.1214/aoms/1177706098 · Cited by 1,122
On orthogonal and symplectic matrix ensembles
Tracy, C. A., & Widom, H. (1996) · Communications in Mathematical Physics · doi:10.1007/bf02099545 · Cited by 786
Random matrices: Universality of local eigenvalue statistics
Tao, T., & Vu, V. (2011) · Acta Mathematica · doi:10.1007/s11511-011-0061-3 · Cited by 442
Rigidity of eigenvalues of generalized Wigner matrices
Erdős, L., Yau, H.-T., & Yin, J. (2012) · Advances in Mathematics · doi:10.1016/j.aim.2011.12.010 · Cited by 345
A Bound on Tail Probabilities for Quadratic Forms in Independent Random Variables
Hanson, D. L., & Wright, F. T. (1971) · The Annals of Mathematical Statistics · doi:10.1214/aoms/1177693335 · Cited by 326
Universality at the Edge of the Spectrum¶in Wigner Random Matrices
Soshnikov, A. (1999) · Communications in Mathematical Physics · doi:10.1007/s002200050743 · Cited by 267
Local Semicircle Law and Complete Delocalization for Wigner Random Matrices
Erdős, L., Schlein, B., & Yau, H.-T. (2009) · Communications in Mathematical Physics · doi:10.1007/s00220-008-0636-9 · Cited by 254
Bulk universality for generalized Wigner matrices
Erdős, L., Yau, H.-T., & Yin, J. (2012) · Probability Theory and Related Fields · doi:10.1007/s00440-011-0390-3 · Cited by 254
Research works printed but not linked: 1
- Erdős, P., Rényi, A.: On the evolution of random graphs. Publ. Math. Inst. Hung. Acad. Sci. 5, 17–61 (1960)
Cited by
Largest eigenvalues of sparse inhomogeneous Erdős–Rényi graphs
Benaych-Georges, F., Bordenave, C., & Knowles, A. (2019) · The Annals of Probability · doi:10.1214/18-aop1293 · Cited by 109
Edge universality for non-Hermitian random matrices
Cipolloni, G., Erdős, L., & Schröder, D. (2020) · Probability Theory and Related Fields · doi:10.1007/s00440-020-01003-7 · Cited by 54
Eigenvector statistics of sparse random matrices
Bourgade, P., Huang, J., & Yau, H. (2017) · Electronic Journal of Probability · doi:10.1214/17-ejp81 · Cited by 48
Singular vector and singular subspace distribution for the matrix denoising model
Bao, Z., Ding, X., & Wang, K. (2021) · The Annals of Statistics · doi:10.1214/20-aos1960 · Cited by 60
Huang, J., Landon, B., & Yau, H. (2020) · The Annals of Probability · doi:10.1214/19-aop1378 · Cited by 65
Metadata
- Authors
- Lee, J. O., & Schnelli, K.
- Affiliations
- Korea Advanced Institute of Science and Technology (KR)
- Institute of Science and Technology Austria (AT)
- Corresponding author
- Lee, Ji Oon
- Venue
- Probability Theory and Related Fields
- Published
- 2017-06-14
- Volume
- 171
- Issue
- 1-2
- Pages
- 543-616
- ISSN
- 0178-8051 (ISSN-L)
- 1432-2064
- OpenAlex
- W2406111469
- Semantic Scholar
- 3d8246f9610ed0952fb2b4b5074c717ec05ebf1b
- Cited by
- 87
- References
- 43
- Citations by year
- 2026: 6 · 2025: 11 · 2024: 11 · 2023: 10 · 2022: 9 · 2021: 9 · 2020: 17 · 2019: 9 · 2018: 1 · 2017: 4
- FWCI
- 5.90
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- Probability Theory and Related Fields (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Samsung Science and Technology Foundation (SSTF-BA1402-04)
- ERC Advanced Grant (RANMAT No. 338804)
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Random Matrices and Applications
- Physical Sciences › Mathematics › Mathematical Physics › Spectral Theory in Mathematical Physics
- Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Quantum chaos and dynamical systems
- Keywords
- sparse random matrices, local legislation, Tracy-Widom distribution, edge universality, Erdős-Rényi random graphs, adjacency matrix, eigenvalue density, extreme eigenvalues, second largest eigenvalue
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-07, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-07 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Local law and Tracy–Widom limit for sparse random matrices. https://app.cochat.ai/papers/10.1007/s00440-017-0787-8
- The paper
- Lee, J. O., & Schnelli, K. (2018). Local law and Tracy–Widom limit for sparse random matrices. Probability Theory and Related Fields. https://doi.org/10.1007/s00440-017-0787-8
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.