Pith. sign in

REVIEW 4 major objections 5 minor 136 references

The sample covariance error has an exact limiting formula in high dimensions.

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 Gaussian data in the proportional limit, the spectral-norm error of the sample covariance converges to γ̂√φ1/(√φ1−√α), with γ̂ solving an equation in the covariance spectrum.

T0 review reviewed 2026-08-02 challenge →

load-bearing objection A plausible exact limit for sample covariance spectral norm error, but the upper-bound derivation contains a concrete duality gap that invalidates the proof as written. the 4 major comments →

arxiv 2607.14460 v1 pith:AD4NWPMT submitted 2026-07-16 math.ST cs.ITmath.ITmath.PRstat.MLstat.TH

Precise sample covariance spectral norm error -- an RDT view

Mihailo Stojnic This is my paper
classification math.ST cs.ITmath.ITmath.PRstat.MLstat.TH MSC 62H1260B2062E20
keywords sample covariance matrixspectral norm errorrandom duality theoryGaussian comparisonproportional high-dimensional asymptoticseffective rankbilinear-quadratic comparisonsample size tradeoff
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 aims to move past scaling-order estimates and determine the exact expected spectral-norm error of the sample covariance matrix for centered Gaussian data. In the proportional regime where n/d tends to a fixed α, it claims the error E∥Σ̂−Σ∥₂ converges to a closed-form expression derived from the true covariance spectrum. The expression reproduces the known isotropic edge when Σ = I, and numerical simulations already track the prediction at dimensions in the thousands. A correct formula of this kind would let practitioners compute the precise benefit of adding samples instead of relying on order-of-magnitude bounds.

Core claim

The paper's central claim is that for centered Gaussian vectors with covariance Σ = U S² Uᵀ, in the limit n/d → α, the expected spectral norm of the sample covariance error converges to the closed-form quantity δ(α) = lim_d γ̂_x √φ₁(γ̂_x) / (√φ₁(γ̂_x) − √α), where φ₁(γ) = (1/d) Σᵢ sᵢ⁴/(γ+sᵢ²)² and γ̂_x solves the stationarity condition lim_d (φ₁(γ̂_x) − φ₂(γ̂_x)φ₃(γ̂_x)) = 0. The author derives this value by sandwiching the error between an RDT upper bound and a new bilinear-quadratic lower bound, then showing the two match in the large-d limit. In the isotropic case Σ = I the formula reduces to the familiar edge 2/√α + 1/α.

What carries the argument

Random duality theory (RDT), the paper's central device, rewrites the error as a maximum over the sphere of a Gaussian process — the random primal ξ(c) — and bounds it from above and below by two Gaussian comparison processes. The upper bound uses a linear 'random dual' L(c); the lower bound introduces a bilinear-quadratic process B(c). A two-replica argument on the overlap q of two copies of the maximization problem is then invoked to show the two bounds coincide. The final formula is organized by the spectral sum φ₁(γ) = (1/d) Σ sᵢ⁴/(γ+sᵢ²)² and the optimal parameter γ̂_x defined by equation (46).

Load-bearing premise

The load-bearing step is the unproved strong-duality swap in equation (26), which equates a non-convex spherical quadratic maximization with its Lagrangian min-max, together with the two-replica condition (82)/(118) that the paper verifies numerically and proves only by contradiction; if either fails, the closed-form limits (48) and (151) collapse.

What would settle it

Choose a deterministic diagonal covariance with a non-trivial spectrum (e.g., s_i equally spaced in [0.5,1]), compute δ(α) from (151) for several α, and compare against high-precision Monte Carlo estimates of E∥Σ̂−Σ∥₂ for d and n around 10,000. A mismatch beyond the expected concentration scale would falsify the claimed equality. A more targeted check evaluates the two-replica condition (118) numerically for a spectrum with two separated eigenvalue clusters; if the inequality reverses, the lower bound no longer matches the upper bound.

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

If this is right

  • If the formula holds, the exact error for any Gaussian covariance spectrum can be obtained by solving the scalar equation (46), without Monte Carlo simulation.
  • It converts the qualitative effective-rank scaling of earlier work into an exact large-d limit, allowing precise statements about how doubling or tripling n changes the error.
  • The isotropic reduction to 2/√α + 1/α ties the result directly to classical random matrix edges.
  • Because the framework is built generically, the same upper/lower comparison strategy is intended to carry over to other covariance error metrics and structured covariance models.

Where Pith is reading between the lines

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

  • One immediate editorial extension: differentiating the closed form with respect to α gives the marginal value of an extra sample, a quantity the paper does not compute explicitly.
  • Remark 4 of the paper already notes that for random covariances with a spectral density the empirical sums can be replaced by integrals; replacing them yields a fully analytic prediction for such priors.
  • A useful stress test is a two-cluster or near-degenerate spectrum, where the two-replica condition (118) rests on numerical verification and the contradiction proof in Theorem 8 is least transparent; agreement there would strengthen confidence in the formula's generality.
  • The matching upper/lower template suggests the same RDT sandwich could produce exact error formulas for related problems such as spiked covariance estimation or covariance estimation under missing data, though the paper leaves those extensions for future work.
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 / 5 minor

Summary. The paper studies the spectral norm error ||Σ̂−Σ||₂ of the sample covariance for centered Gaussian vectors with covariance Σ in the proportional regime n/d→α. It develops a Random Duality Theory (RDT) framework: a Gaussian-process upper bound for Eξ(c), an explicit saddle-point evaluation of the dual, and a new bilinear-quadratic lower-bound mechanism. The main claim, Theorem 11 (Eq. 151), is that the limiting expected spectral norm equals δ(α)=lim_d γ̂_x √φ₁(γ̂_x)/(√φ₁(γ̂_x)−√α), where γ̂_x solves the fixed-point equation (46) and φ₁ is defined in (43). For Σ=I the formula reduces to 2/√α+1/α, matching the known Wishart edge. The paper also reports numerical simulations for d up to a few thousand that agree well with the formula.

Significance. If the result is correct, this is a valuable precise characterization of sample-covariance error beyond scaling laws, with a deterministic, parameter-free formula and a proof strategy (RDT + replica lower bounds) that could potentially be exported to other problems. The claimed reduction to the known Wishart edge is a useful consistency check, and the simulations are encouraging. However, the proof as written has several load-bearing gaps, most seriously an incorrect unconstrained dual evaluation (Eq. 30) and an unproved strong-duality step (Eq. 26). These issues directly affect the central claim, so the paper cannot be accepted in its present form.

major comments (4)
  1. [Section 3.3, Eq. (30)] The closed-form evaluation of L(c) is false as stated. Eq. (30) minimizes over γ_x without any dual-feasibility domain. Take d=2, S=diag(1,2), c²=9/4, g=(1,0). The primal problem is max x₁ subject to x₁²+x₂²=1 and x₁²+4x₂²=9/4, whose value is √(7/12)=0.7638. The RHS of (30) is inf_{γ_x} √((γ_x+9/4)/(γ_x+1)) over real γ_x with nonnegative radicand, which is 0, attained at γ_x=−9/4. Equality would only hold under an additional restriction such as γ_x+s_i²≤0 (here γ_x≤−4), which is never stated. Since Theorems 3 and 11 inherit this step, the upper-bound proof and the exact-limit proof are not valid as written.
  2. [Section 3.3, Eq. (26)] Equation (26) asserts strong duality between the QCQP L(c)=max_{‖x‖=1,‖Sx‖=c} g^T Sx and its Lagrangian dual. The constraints are two nonconvex quadratic equalities, and no Slater-type, S-procedure, or coercivity argument is provided. For such problems a positive duality gap is possible, and every subsequent closed-form expression for L(c), including (30) and the final limit (151), depends on this swap. The paper needs a rigorous justification of (26) or a different derivation of L(c).
  3. [Remark 1; Eqs. (23), (33)] The paper repeatedly uses concentration and interchanges of E with max_c and with lim_d. Eq. (23) writes Eλ_n(Σ̂−Σ)=max_c((Eξ)²−c²) based on the assertion in Remark 1 that all objects 'trivially concentrate.' Uniform concentration over the compact c-domain is a nontrivial ingredient that is not proved. Likewise, (31)–(33) pass limits through the max over c and min over γ_x. These interchanges are load-bearing for the upper bound and hence for Theorem 11; a rigorous treatment with quantitative tail bounds is needed.
  4. [Section 3.4.2 and Theorem 8] The lower-bound matching step relies on the flatness implication imported from [120,121], and the verification of condition (118) is not fully rigorous. Theorem 8's proof states 'from (39), one also has γ̃_x ≤ 2c²' but (39) actually gives γ̃_x≤0 and γ̃_x+2c²≤0, i.e., γ̃_x≤−2c²; the subsequent claim γ̃_x≠−c² requires this corrected inequality and additional justification. More importantly, condition (118) is a global inequality over t∈(0,1) and q∈(−1,1), and the stationary-point contradiction in Theorem 8 does not address all possible boundary or infimum cases. The numerical check in Figure 1 for one spectrum is suggestive but does not constitute a proof for general Σ.
minor comments (5)
  1. [General notation] The paper uses m and n interchangeably in several places (e.g., the proof of Theorem 1 and eq. (64)), and 'y∈S^m' appears where S^n is intended.
  2. [Eq. (64)] There is a typo '1‘/2c²' in the expression for EG_u(X^(a1))G_u(X^(a2)); the correct factor should be 1/(2c²).
  3. [Section 3.6, Figures 2–3] The simulation section does not report the number of Monte-Carlo trials, error bars, or the exact simulation protocol. This makes it hard to assess the claimed 'excellent agreement.'
  4. [Theorem 8 proof] The parenthetical 'from (39), one also has γ̃_x ≤ 2c²' appears to be a sign typo; (39) implies γ̃_x ≤ −2c². Please correct and clarify the implication for γ̃_x ≠ −c².
  5. [Section 3.4.2.4] The statement 'We tested quite a few ensembles and always obtained that (118) holds' is informal. If this is only numerical evidence, it should be clearly labeled as such and not used as a substitute for a proof.

Circularity Check

0 steps flagged

No circularity: the claimed limit comes from a deterministic fixed-point equation; RDT self-citations are contextual, not load-bearing.

full rationale

The paper's central prediction δ̂u(α) is not an input recycled as an output: it is computed from the deterministic fixed-point equation (46), whose ingredients φ1, φ2, φ3 and γ̂x depend only on the covariance spectrum {s_i} and α, and no parameter is fitted to E∥Σ̂−Σ∥ or to the simulation values. The upper bound (48)-(49) is derived from explicit Slepian/Gordon comparisons (Theorems 1-2, citing [45,108]) and Lagrangian calculus; the lower-bound half does not simply assert equality but makes it contingent on condition (118), which Theorem 8 verifies analytically (with Figure 1 as numerical corroboration). Citations to the author's RDT program [111-118] appear as methodological framing and as generalizations of external comparison inequalities, but the load-bearing comparison theorems are attributed to Slepian [108], Gordon [45], and Talagrand [120,121]; moreover, the paper supplies its own verification of the replica-system condition rather than importing the conclusion. The isotropic limit (59) and Figures 2-3 provide external benchmarks. The skeptic's objection — that Eq. (26) invokes unproved strong duality and that Eq. (30) omits the dual-feasibility domain of γx — is an omitted-justification/correctness gap in the upper-bound proof, not a circular reduction of the claimed limit to its own inputs; it would invalidate the proof if correct but does not make the derivation definitionally circular. No fitted-input-as-prediction, uniqueness-imported-by-self-citation, or renaming step is present.

Axiom & Free-Parameter Ledger

0 free parameters · 6 axioms · 0 invented entities

The paper introduces no fitted parameters or new physical entities. The 'bilinear-quadratic mechanism' and '2-replica system' are mathematical techniques, not independently falsifiable objects. The main axioms are standard Gaussian comparison theorems, Talagrand's replica machinery, and two ad hoc analytic shortcuts: strong duality for a non-convex QCQP and unproved concentration.

axioms (6)
  • standard math Slepian/Gordon comparison theorems as stated in Theorem 2 and Theorem 10 (from [45,108])
    Used in Theorems 1,4,5,6,9 to compare Gaussian processes; the direction of the inequality is load-bearing.
  • standard math Talagrand's 2-replica flatness machinery for spherical models
    In Section 3.4.2.2, condition (118) and eq. (82)-(83) invoke results from [120,121] that the no-double-free-energy condition implies ED(c,1)=ED(c,0); the paper does not reprove this implication.
  • ad hoc to paper Strong duality for the QCQP defining L(c)
    Eq. (26) in Section 3.3 asserts min_x max_γ L = max_γ min_x L for a non-convex quadratic program with two quadratic constraints; no justification is given.
  • ad hoc to paper Concentration of ξ(c), L(c), B(c) and interchange of E and max_c
    Remark 1 states all objects concentrate and the paper does not supply bounds; used in (23), (33), (135)-(136), and Theorem 11.
  • domain assumption Eigenvalues of S are positive and lie in a fixed interval independent of d
    Section 2, paragraph after eq. (10): 'we assume that the eigenvalues of S are positive and belong to an interval independent of d'.
  • domain assumption Gaussianity and proportional scaling n/d = α fixed
    Problem setup eqs (1)-(3).

reviewed 2026-08-02 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Precise sample covariance spectral norm error -- an RDT view." pith.science (2026). https://pith.science/paper/AD4NWPMT

@misc{pith2026260714460,
  author       = {Pith},
  title        = {Pith review of: Precise sample covariance spectral norm error -- an RDT view},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AD4NWPMT}},
  note         = {Machine review of arXiv:2607.14460}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We study the sample covariance error of centered Gaussians. A remarkable breakthrough [66] established the correct error scaling order and explicitly revealed the critical role of both the effective rank and the true covariance spectrum. In this work, we move beyond scaling characterizations and determine the precise limiting value of the error's spectral norm. To do so, we develop a generic framework based on Random Duality Theory (RDT). Within this framework, we first determine closed-form, explicit RDT-based upper bounds. We then establish complementary lower bounds by introducing a novel bilinear-quadratic RDT lower-bounding mechanism. By combining this mechanism with a two-replica systems bounding strategy, we show that our lower and upper bounds match in large-dimensional contexts. Our theoretical results are supplemented with numerical evaluations and simulations, demonstrating an excellent agreement already for problem sizes on the order of thousands.

Figures

Figures reproduced from arXiv: 2607.14460 by Mihailo Stojnic.

Figure 1
Figure 1. Figure 1: √ minγx,νx L¯(2) 2 as a function of q for varying t; s = linspace[0.5, 1]; d = 3000 and for which the following holds [PITH_FULL_IMAGE:figures/full_fig_p022_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Sample covariance error, δn = δαd = E∥Σˆ − Σ∥2, as a function of d; α = 1, i.e., n = d; s = linspaced [0.5, 1] The conducted analysis allows to obtain very precise estimation error characterizations. Consequently, it enables to accurately predict concrete effects of the increased number of samples. We show this in [PITH_FULL_IMAGE:figures/full_fig_p028_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Sample covariance error, δn = δαd = E∥Σˆ − Σ∥2, as a function of d; varying sample complexity n, i.e., varying α = n d ; s = linspaced [0.5, 1] 4 Conclusion We studied the sample covariance error of centered Gaussians. To move beyond scaling characterizations and determine the precise limiting value of the error’s spectral norm, we have developed a generic framework based on Random Duality Theory (RDT). Th… 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

136 extracted references · 7 linked inside Pith

  1. [1]

    P. Abdalla. Covariance estimation under missing observations andl4 −l 2 moment equivalence.Elec- tronic Journal of Statistics, 18(1):2057–2108, 2024

    2057

  2. [2]

    Abdalla and N Zhivotovskiy

    P. Abdalla and N Zhivotovskiy. Covariance estimation: optimal dimension-free guarantees for adver- sarial corruption and heavy tails.Journal of the European Mathematical Society, 28(4):1809–1847, 2026

    2026

  3. [3]

    Adamczak

    R. Adamczak. A note on the Hanson-Wright inequality for random vectors with dependencies.Elec- tronic Communications in Probability, 20:1–13, 2015

    2015

  4. [4]

    Adamczak, A

    R. Adamczak, A. E. Litvak, A. Pajor, and N. Tomczak-Jaegermann. Quantitative estimates of the convergence of the empirical covariance matrix in log-concave ensembles.Journal of the American Mathematical Society, 23(2):535–561, 2010

    2010

  5. [5]

    Adamczak, A

    R. Adamczak, A. E. Litvak, A. Pajor, and N. Tomczak-Jaegermann. Sharp bounds on the rate of convergence of the empirical covariance matrix.Comptes Rendus Mathématique, 349(3–4):195–200, 2011

    2011

  6. [6]

    Agostinelli, A

    C. Agostinelli, A. Leung, and K. Yu. Robust low-rank covariance matrix estimation with a general pattern of missing values.Signal Processing, 194:108433, 2022

    2022

  7. [7]

    Al-Ghattas, J

    O. Al-Ghattas, J. Chen, and D. Sanz-Alonso. Sharp concentration of simple random tensors.Infor- mation and Inference: A Journal of the IMA, 14(4):iaaf029, 2025

    2025

  8. [8]

    Bai and S

    J. Bai and S. Shi. Estimating high dimensional covariance matrices and its applications.Annals of Economics and Finance, 12(2):199–215, 2011

    2011

  9. [9]

    Bai and J

    Z. Bai and J. W. Silverstein.Spectral Analysis of Large Dimensional Random Matrices. Springer Series in Statistics. Springer, 2010

    2010

  10. [10]

    J. Baik, G. Ben Arous, and S. Peche. Phase transition of the largest eigenvalue for non-null complex sample covariance matrices.The Annals of Probability, 33(5):1643–1697, 2005

    2005

  11. [11]

    A. S. Bandeira, G. Cipolloni, D. Schroder, and R. van Handel. Matrix concentration inequalities and free probability ii. Two-sided bounds and applications. 2024. available online athttp://arxiv.org/ abs/2406.11453

    arXiv 2024

  12. [12]

    A. S. Bandeira and R. van Handel. Sharp nonasymptotic bounds on the norm of random matrices with independent entries.The Annals of Probability, 44(4):2479–2506, 2016

    2016

  13. [13]

    Bandeira, M.T

    A.S. Bandeira, M.T. Boedihardjo, and R. van Handel. Matrix concentration inequalities and free probability.Inventiones Mathematicae, 234:419–487, 2023

    2023

  14. [14]

    Barbier, N

    J. Barbier, N. Macris, and L. Miolane. The layered structure of tensor estimation and its mutual information. In2017 55th Annual Allerton Conference on Communication, Control, and Computing (Allerton), pages 1056–1063. IEEE, 2017

    2017

  15. [15]

    W. Bednorz. Concentration via chaining method and its applications. 2014. available online at http://arxiv.org/abs/1405.0676. 29

    Pith/arXiv arXiv 2014

  16. [16]

    W. Bednorz. Bounds for stochastic processes on product index spaces. InHigh Dimensional Probability VII: The Cargese Volume, pages 327–357. Birkhauser / Springer International Publishing, 2016

    2016

  17. [17]

    J. K. Behne and G. Reeves. Fundamental limits for rank-one matrix estimation with groupwise het- eroskedasticity. InProceedings of The 25th International Conference on Artificial Intelligence and Statistics, volume 151 ofProceedings of Machine Learning Research, pages 8650–8672. PMLR, 2022

    2022

  18. [18]

    Benaych-Georges, A

    F. Benaych-Georges, A. Guionnet, and M. Maida. Large deviations of the extreme eigenvalues of random deformations of matrices.Probab. Theory Relat. Fields, 154:703–751, 2012

    2012

  19. [19]

    P. J. Bickel and E. Levina. Covariance regularization by thresholding.The Annals of Statistics, 36(6):2577–2604, 2008

    2008

  20. [20]

    Biroli and A

    G. Biroli and A. Guionnet. Large deviations for the largest eigenvalues and eigenvectors of spiked Gaussian random matrices.Electronic Communications in Probability, 25(none):1 – 13, 2020

    2020

  21. [21]

    Bourgain

    J. Bourgain. Random points in isotropic convex sets. InConvex Geometric Analysis (Berkeley, CA, 1996), volume 34 ofMathematical Sciences Research Institute Publications, pages 53–58. Cambridge University Press, 1999

    1996

  22. [22]

    Brailovskaya and R

    T. Brailovskaya and R. van Handel. Universality and sharp matrix concentration inequalities.Geo- metric and Functional Analysis, 34:1003–1045, 2024

    2024

  23. [23]

    Bunea and L

    F. Bunea and L. Xiao. On the sample covariance matrix estimator of reduced effective rank population matrices, with applications to fpca.Bernoulli, 21(2):1200–1230, 2015

    2015

  24. [24]

    T. T. Cai, R. Han, and A. R. Zhang. On the non-asymptotic concentration of heteroskedastic Wishart- type matrix.Electronic Journal of Probability, 27:1–40, 2022

    2022

  25. [25]

    T. T. Cai, Z. Ren, and H. H. Zhou. Optimal rates of convergence for estimating Toeplitz covariance matrices.Probability Theory and Related Fields, 156(1-2):101–143, 2013

    2013

  26. [26]

    T. T. Cai, Z. Ren, and H. H Zhou. Estimating structured high-dimensional covariance and precision matrices: Optimal rates and adaptive estimation.Statistical Science, 31(1):67–81, 2016

    2016

  27. [27]

    T. T. Cai and A. Zhang. Minimax rate-optimal estimation of high-dimensional covariance matrices with incomplete data.Journal of Multivariate Analysis, 150:55–74, 2016

    2016

  28. [28]

    T. T. Cai and A. Zhang. Optimal estimation of high-dimensional sparse covariance matrices with missing data.Communications in Statistics-Theory and Methods, pages 1–24, 2024

    2024

  29. [29]

    T. T. Cai and H. H. Zhou. Optimal rates of convergence for sparse covariance matrix estimation.The Annals of Statistics, 40(5):2389–2420, 2012

    2012

  30. [30]

    Chen and D

    J. Chen and D. Sanz-Alonso. Concentration inequalities for sample cross-covariances. 2026. available online athttp://arxiv.org/abs/2605.16733

    Pith/arXiv arXiv 2026

  31. [31]

    Chen, J.Y

    K.X. Chen, J.Y. Ren, X.J. Wu, and J. Kittler. Covariance descriptors on a Gaussian manifold and their application to image set classification.Pattern Recognition, 106:107463, 2020

    2020

  32. [32]

    M. Chen, C. Gao, and Z. Ren. Robust covariance and scatter matrix estimation under Huber’s contamination model.The Annals of Statistics, 46(5):1932–1960, 2018

    1932

  33. [33]

    Dahmen, D

    J. Dahmen, D. Keysers, M. Pitz, and H. Ney. Structured covariance matrices for statistical image object recognition. InMustererkennung 2000, 22. DAGM-Symposium, Kiel, September 2000, pages 99–106. Springer, 2000

    2000

  34. [34]

    A. S. Dalalyan and A. Minasyan. All-in-one robust estimator of the Gaussian mean.The Annals of Statistics, 50(2):1193–1219, 2022. 30

    2022

  35. [35]

    Diakonikolas and D

    I. Diakonikolas and D. M. Kane. Implicit high-order moment tensor estimation and learning latent variable models. In2025 IEEE 66th Annual Symposium on Foundations of Computer Science (FOCS), pages 1228–1247, 2025

    2025

  36. [36]

    Diakonikolas, D

    I. Diakonikolas, D. M. Kane, and A. Pensia. Outlier robust mean estimation with subgaussian rates via stability. InAdvances in Neural Information Processing Systems, volume 33, pages 18398–18408, 2020

    2020

  37. [37]

    S. Dirksen. Tail bounds via generic chaining.Electronic Journal of Probability, 20(53):1–29, 2015

    2015

  38. [38]

    D. L. Donoho, M. Gavish, and I. M. Johnstone. Optimal shrinkage of eigenvalues in the spiked covariance model.The Annals of Statistics, 46(4):1742–1778, 2018

    2018

  39. [39]

    Z. Dou, Z. Fan, and H. H. Zhou. Rates of estimation for high-dimensional multi-reference alignment. The Annals of Statistics, 52(1):1–32, 2024

    2024

  40. [40]

    R. Engle. Dynamic conditional correlation: A simple class of multivariate generalized autoregressive conditional heteroskedasticity models.Journal of Business & Economic Statistics, 20(3):337–350, 2002

    2002

  41. [41]

    J. Fan, P. Rigollet, and W. Wang. Estimation of functionals of sparse covariance matrices.The Annals of Statistics, 43(6):2616–2646, 2015

    2015

  42. [42]

    Friedman, T

    J. Friedman, T. Hastie, and R. Tibshirani. Sparse inverse covariance estimation with the graphical lasso.Biostatistics, 9(3):432–441, 2008

    2008

  43. [43]

    Giannopoulos, M

    A. Giannopoulos, M. Hartzoulaki, and A. Tsolomitis. Random points in isotropic unconditional convex bodies.Journal of the London Mathematical Society, 72(3):779–798, 2005

    2005

  44. [44]

    I. Giulini. Robust dimension-free gram operator estimates.Bernoulli, 24(4B):3864–3923, 2018

    2018

  45. [45]

    Y. Gordon. Some inequalities for Gaussian processes and applications.Israel Journal of Mathematics, 50(4):265–289, 1985

    1985

  46. [46]

    Guionnet, J

    A. Guionnet, J. Ko, F. Krzakala, and L. Zdeborová. Low-rank matrix estimation with inhomogeneous noise.Information and Inference: A Journal of the IMA, 14(2):iaaf010, 06 2025

    2025

  47. [47]

    Haghighatshoar and G

    S. Haghighatshoar and G. Caire. Low-complexity massive MIMO subspace estimation and tracking from low-dimensional projections.IEEE Transactions on Signal Processing, 65(2):303–318, 2017

    2017

  48. [48]

    F. R. Hampel, E. M. Ronchetti, P. J. Rousseeuw, and W. A. Stahel.Robust Statistics: The Approach Based on Influence Functions. Wiley Series in Probability and Mathematical Statistics. John Wiley & Sons, New York, 1986

    1986

  49. [49]

    Q. Han. Exact bounds for some quadratic empirical processes with applications, 2024. available online athttp://arxiv.org/abs/2207.13594

    Pith/arXiv arXiv 2024

  50. [50]

    R. Han, R. Willett, and A. R. Zhang. An optimal statistical and computational framework for gener- alized tensor estimation.The Annals of Statistics, 50(1):293–319, 2022

    2022

  51. [51]

    Han and W

    Y. Han and W. B. Wu. Test for high dimensional covariance matrices.The Annals of Statistics, 48(6):3565–3588, 2020

    2020

  52. [52]

    A. O. Hero and B. Rajaratnam. Hub discovery in partial correlation graphs.IEEE Transactions on Information Theory, 58(9):6064–6078, 2012

    2012

  53. [53]

    Holtz.Sparse Grid Quadrature in High Dimensions with Applications in Finance and Insurance, volume 77 ofLecture Notes in Computational Science and Engineering

    M. Holtz.Sparse Grid Quadrature in High Dimensions with Applications in Finance and Insurance, volume 77 ofLecture Notes in Computational Science and Engineering. Springer Science & Business Media, 2011

    2011

  54. [54]

    P. J. Huber. Robust estimation of a location parameter.The Annals of Mathematical Statistics, 35(1):73–101, 1964. 31

    1964

  55. [55]

    P. J. Huber.Robust Statistics. Wiley Series in Probability and Mathematical Statistics. John Wiley & Sons, New York, 1981

    1981

  56. [56]

    Husson and B

    J. Husson and B. McKenna. Large deviations for the largest eigenvalue of generalized sample covariance matrices.Electronic Journal of Probability, 29:1–48, 2024

    2024

  57. [57]

    Kannan, L

    R. Kannan, L. Lovasz, and M. Simonovits. Random walks and ano∗(n5)volume algorithm for convex bodies.Random Structures & Algorithms, 11(1):1–50, 1997

    1997

  58. [58]

    El Karoui

    N. El Karoui. Operator norm consistent estimation of large-dimensional sparse covariance matrices. The Annals of Statistics, 36(6):2717–2756, December 2008

    2008

  59. [59]

    Y. Ke, S. Minsker, Z. Ren, Q. Sun, and W.-X. Zhou. User-friendly covariance estimation for heavy- tailed distributions.Statistical Science, 34(3):454–471, 2019

    2019

  60. [60]

    M. B. Khalilsarai, T. Yang, S. Haghighatshoar, and G. Caire. Structured channel covariance estimation from limited samples in massive MIMO. InIEEE International Conference on Communications (ICC), pages 1–7, 2020

    2020

  61. [61]

    Klartag and S

    B. Klartag and S. Mendelson. Empirical processes and random projections.Journal of Functional Analysis, 225(1):229–245, 2005

    2005

  62. [62]

    Koltchinskii

    V. Koltchinskii. Asymptotic efficiency in high-dimensional covariance estimation. InProceedings of the International Congress of Mathematicians (ICM 2018), page 2921. World Scientific, 2018

    2018

  63. [63]

    Koltchinskii

    V. Koltchinskii. Efficient estimation of smooth functionals in Gaussian shift models.Annales de l’Institut Henri Poincare, Probabilites et Statistiques, 57(1):1–31, 2021

    2021

  64. [64]

    Koltchinskii

    V. Koltchinskii. Estimation of smooth functionals in high-dimensional models: Bootstrap chains and Gaussian approximation.The Annals of Statistics, 50(4):2386–2415, 2022

    2022

  65. [65]

    Koltchinskii, M

    V. Koltchinskii, M. Loffler, and R. Nickl. Efficient estimation of linear functionals of principal compo- nents.The Annals of Statistics, 48(1):464–490, 2020

    2020

  66. [66]

    Koltchinskii and K

    V. Koltchinskii and K. Lounici. Concentration inequalities and moment bounds for sample covariance operators.Bernoulli, 23(1):110–133, 2017

    2017

  67. [67]

    Koltchinskii and K

    V. Koltchinskii and K. Lounici. New asymptotic results in principal component analysis.Sankhya A, 79(2):254–297, 2017

    2017

  68. [68]

    Koltchinskii and K

    V. Koltchinskii and K. Lounici. Normal approximation and concentration of spectral projectors of sample covariance.The Annals of Statistics, 45(1):121–157, 2017

    2017

  69. [69]

    Koltchinskii and M

    V. Koltchinskii and M. Zhilova. Estimation of smooth functionals in normal models: bias reduction and asymptotic efficiency.The Annals of Statistics, 49(5):2847–2873, 2021

    2021

  70. [70]

    Krim and M

    H. Krim and M. Viberg. Two decades of array signal processing research: the parametric approach. IEEE Signal Processing Magazine, 13(4):67–94, 1996

    1996

  71. [71]

    K. A. Lai, A. B. Rao, and S. Vempala. Agnostic estimation of mean and covariance. InIEEE 57th Annual Symposium on Foundations of Computer Science (FOCS), pages 665–674. IEEE, 2016

    2016

  72. [72]

    Langfelder and S

    P. Langfelder and S. Horvath. Wgcna: an r package for weighted correlation network analysis.BMC bioinformatics, 9:1–13, 2008

    2008

  73. [73]

    Latala, R

    R. Latala, R. van Handel, and P. Youssef. The dimension-free structure of nonhomogeneous random matrices.Inventiones Mathematicae, 214(2):1031–1080, 2018

    2018

  74. [74]

    Ledoit and M

    O. Ledoit and M. Wolf. Some hypothesis tests for the covariance matrix when the dimension is large compared to the sample size.The Annals of Statistics, 30(4):1081–1102, 2002. 32

    2002

  75. [75]

    Ledoit and M

    O. Ledoit and M. Wolf. Improved estimation of the covariance matrix of stock returns with an appli- cation to portfolio selection.Journal of Empirical Finance, 10(5):603–621, 2003

    2003

  76. [76]

    Awell-conditionedestimatorforlarge-dimensionalcovariancematrices.Journal of Multivariate Analysis, 88(2):365–411, 2004

    O.LedoitandM.Wolf. Awell-conditionedestimatorforlarge-dimensionalcovariancematrices.Journal of Multivariate Analysis, 88(2):365–411, 2004

    2004

  77. [77]

    Leng and G

    C. Leng and G. Pan. Covariance estimation via sparse Kronecker structures.Bernoulli, 24(4B):3833– 3863, 2018

    2018

  78. [78]

    Lesieur, L

    T. Lesieur, L. Miolane, M. Lelarge, F. Krzakala, and L. Zdeborová. Statistical and computational phase transitions in spiked tensor estimation. In2017 IEEE International Symposium on Information Theory (ISIT), pages 511–515, 2017

    2017

  79. [79]

    K. Lounici. High-dimensional covariance matrix estimation with missing observations.Bernoulli, 20(3):1029–1058, 2014

    2014

  80. [80]

    Lounici and G

    K. Lounici and G. Pacreau. Robust covariance estimation with missing values and cell-wise contami- nation. InAdvances in Neural Information Processing Systems, volume 36, pages 72124–72136, 2023

    2023

Showing first 80 references.

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