Pith. sign in

REVIEW 4 major objections 5 minor 38 references

A known linear restriction on the regression coefficients adds residual degrees of freedom that make a Tyler-based covariance shrinker tail-free and let a Stein combination beat the unrestricted estimator.

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

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

T0 review · deepseek-v4-flash

2026-08-01 11:34 UTC pith:52H7UFQA

load-bearing objection A genuinely new restricted-shrinkage idea whose main theorems are unproven as printed; the union-bound gap in Theorem 1 is real and load-bearing. the 4 major comments →

arxiv 2607.20576 v2 pith:52H7UFQA submitted 2026-07-22 stat.ME math.STstat.TH

Restricted nonlinear shrinkage of high-dimensional residual covariance matrices in multivariate regressions

Hamid Karamikabir (1) , Mohammad Arashi (2) ((1) Department of Statistics , Faculty of Intelligent Systems Engineering , Data Science , Persian Gulf University , Bushehr , Iran , (2) Department of Statistics
show 4 more authors
Faculty of Mathematical Sciences Ferdowsi University of Mashhad Mashhad Iran)
This is my paper
classification stat.ME math.STstat.TH MSC 62H1262J0560B2062C20
keywords analytic shrinkageelliptical distributionhigh-dimensional covarianceMarchenko-Pastur lawmultivariate regressionrestricted estimationStein estimationTyler's M-estimator
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 paper studies estimating the p×p residual covariance of a multivariate regression in the high-dimensional regime where p and n grow together. Its central claim is that when the coefficient matrix satisfies a known linear restriction of rank q, the restricted residuals carry q extra degrees of freedom, and this can be converted into a strictly better covariance estimator. The key move is to shrink a scale-invariant scatter, Tyler's M-estimator, rather than the sample residual covariance, because the scatter's limiting spectrum is identical for all elliptical error distributions with finite second moment, while the sample covariance's spectrum drifts with the error tail. At the reduced effective aspect ratio p/(n−d+q), the restricted shrinker attains the best rotation-equivariant estimator, and a positive-part Stein-type combination dominates the unrestricted shrinker when q≥3 and degrades gracefully if the restriction is misspecified.

Core claim

On its own terms, the paper establishes two main results. Theorem 1 shows that the empirical spectral distribution of Tyler's M-estimator formed from the restricted residuals converges to the same generalized Marchenko–Pastur law as under Gaussian errors, determined only by the effective aspect ratio c̃ = p/(n−d+q) and the population spectrum of the shape matrix, independent of the error radii. Consequently, the analytic nonlinear shrinkage of this scatter is consistent for the rotation-equivariant oracle. Theorem 2 shows that, for q≥3 and a local misspecification of the restriction, the positive-part Stein estimator bΣS = (1−κ)bΣURE + κbΣRRE has risk R(bΣS) ≤ R(bΣURE) − κ⋆(q−2)²(c_n−c̃_n)²p

What carries the argument

The construction rests on Tyler's M-estimator of scatter, the unique trace-normalized fixed point of bV = (p/m)Σᵢ rᵢrᵢᵀ/(rᵢᵀ bV⁻¹ rᵢ), which depends on the residuals only through their directions and is therefore invariant to per-row scales. Shrinking this scatter with the analytic nonlinear shrinkage operator φ_c, instead of shrinking the residual sample covariance, makes the limiting spectrum distribution-free over the elliptical family. The reduced effective aspect ratio c̃ = p/(n−d+q) comes from the projection identity bEr = (In−Pr)E with rank m = n−d+q; the final estimator is a convex Stein-type combination whose intensity κn = min{1, (q−2)+/((n−d)Tn)} interpolates between the restricte

Load-bearing premise

The load-bearing premise is that the least-squares projection makes the restricted residual directions asymptotically match the true error directions uniformly, which requires vanishing leverage (maxᵢ(Pr)ᵢᵢ → 0) and finite second moments of the error radii and their inverses; if high-leverage rows persist or E(R²) is infinite, the tail-free spectrum result and all downstream oracle and dominance claims can fail.

What would settle it

Simulate a multivariate regression in the proportional regime with a designed high-leverage row whose leverage does not vanish, and draw errors from a heavy-tailed elliptical law with infinite second moment (e.g., multivariate t with 2 degrees of freedom). Compute the empirical spectral distribution of the restricted Tyler scatter: if its bulk depends on the error tail or deviates from the generalized Marchenko–Pastur law at c̃ = p/(n−d+q), Theorem 1 is refuted.

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

If this is right

  • The restricted robust shrinker attains the rotation-equivariant oracle: L(bΣRRE)/L⋆(c̃,H) →p 1, so no rotation-equivariant estimator can beat it asymptotically at the effective aspect ratio.
  • The restricted oracle risk is smaller than the unrestricted one by the factor (n−d)/(n−d+q) = (1−γ)/(1−γ+τ), making the efficiency gain a direct function of the extra residual degrees of freedom.
  • When q≥3 and the restriction is locally misspecified, the positive-part Stein estimator strictly dominates the unrestricted analytic shrinker, with a risk gap of order (q−2)²(q/n)²·p/(n−d)².
  • Under arbitrary misspecification of the restriction, the Stein estimator satisfies R(bΣS) ≤ R(bΣURE) + C/n, so it never loses more than O(n⁻¹) even when the restriction is grossly wrong.
  • When the restriction is selected from the data, the scale-corrected version of the restricted estimator is asymptotically unbiased and retains the known-restriction efficiency, at negligible cost compared to sample splitting.

Where Pith is reading between the lines

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

  • Editorial extension: the distribution-free spectrum result is driven by direction-based scatter, so other scale-invariant estimators (such as spatial-sign covariance) could plausibly replace Tyler's estimator in this construction, possibly extending the approach to aspect ratios at or above one.
  • Editorial extension: because the gain grows with the rank q of the restriction and the Stein rule protects against misspecification, a data-driven choice of restriction matrix could be framed as a covariance-estimation problem: select the largest q that survives a contrast test, and let the Stein combination insure against selecting too many.
  • The paper itself flags two soft spots: the q−2 intensity in the Stein rule is described as definitional rather than derived by the divergence, and the post-selection correction fixes only the scale, leaving second-order shape effects to future work. The dominance claim should be read as conditional on those two choices.
  • Editorial extension: the proof of Theorem 2 uses the Gaussian scale-mixture representation of elliptical errors and asymptotic independence of the residual scatter from the fitted contrasts; for non-elliptical or dependent errors, the dominance rate of Theorem 2 would need separate verification, even if the distribution-free spectrum result of Theorem 1 remains plausible.

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 proposes restricted nonlinear shrinkage estimators for the residual covariance/scatter matrix in high-dimensional multivariate regression under a known linear restriction on the coefficient matrix. The construction uses Tyler's M-estimator of the restricted residuals, applies Ledoit–Wolf analytic nonlinear shrinkage at the reduced effective aspect ratio c~=p/(n−d+q), and forms a positive-part Stein-type combination with the unrestricted shrinker. The main claims are: Theorem 1, that the empirical spectral distribution of the restricted Tyler scatter converges to a generalized Marchenko–Pastur law independent of the radial law; Theorem 2, that the positive-part Stein estimator asymptotically dominates the unrestricted analytic shrinker under a local misspecification; Theorems 3–4, oracle attainment and rotation-equivariant optimality; and Theorem 5, robustness to misspecified restrictions. The paper includes simulations, a designed growth-curve experiment, and two real-data analyses.

Significance. If the results hold, the paper makes a useful contribution: it identifies linear restrictions on the coefficient matrix as a structural source of information for residual covariance estimation, extends distribution-free properties of Tyler's scatter to regression residuals under mild moment conditions, and offers a practical, positive-semidefinite convex-combination estimator with a safety guarantee under misspecification. The manuscript ships reproducible code and contains extensive simulations and real-data illustrations, which are strengths. However, the central proofs as printed are not verifiable. The appendix repeatedly cites non-existent supplementary sections (S1, S3, S5, S6, S8), a key uniform direction-match step in Theorem 1 is not established from the stated conditions, the asymptotic-independence step in Theorem 2 is asserted rather than derived, and the scale calibrator bσ² is defined through an undefined 'population analogue.' These are load-bearing gaps, not mere presentation issues.

major comments (4)
  1. [Appendix, Proof of Theorem 1, Step 2] The proof asserts max_i ||∑_{j≠i}(Pr)_{ij} e_j||/||e_i|| = o_p(1) by bounding the numerator's second moment by μ_2 h_n and claiming E(R_i^{-2})<∞ makes R_i^{-1}=O_p(1) uniformly by a union bound. This is not valid as stated. Markov's inequality gives O_p(h_n^{1/2}) only for fixed i; a union bound over i requires n h_n→0, whereas Condition 2 states only h_n→0 and permits h_n≍(d−q)/n with d−q=n^α, 0<α<1. Moreover E(R^{-2})<∞ alone does not give max_i R_i^{-1}=O_p(1); for lower-tail densities P(R<ε)≍ε^2 the maximum inverse radius is typically O_p(n^{1/2}), not O_p(1). Thus the direction-match premise is not a consequence of Conditions 1–4. This is load-bearing: it is the only step replacing restricted residual directions by error directions; without it, the distribution-free MP law at c~=p/(n−d+q), oracle consistency of bΣ_RRE, and Theorem 2's dominance are unsupported.
  2. [Appendix passim; missing supplementary sections] The appendix repeatedly cites 'Section S1,' 'Section S3,' 'Section S5,' 'Section S6,' and 'Section S8' (e.g., Proof of Theorem 1 Step 0; Proof of Theorem 2; Proof of Theorem 5; Proof of Proposition 2), but no supplementary file is included. These references carry load-bearing derivations: the effective-row reduction and leverage identity (S1), the weighted law-of-large-numbers and Hanson–Wright concentration (S3), the bulk eigenvector alignment (S6), and the oracle elasticity/monotonicity (S8). Without them, Theorems 2 and 5 and Proposition 2 are not checkable. The authors must either include the supplement or fold the necessary arguments into the main text before the manuscript can be evaluated.
  3. [Proof of Theorem 2, asymptotic-independence step] The proof states that bΣ_r and R bB are 'asymptotically independent' after showing (I_n−P_r)M = M and ||M||_F^2 = tr(Q) = O(q/n). The argument 'by Isserlis' theorem all odd joint cumulants vanish' only establishes uncorrelatedness between linear and quadratic terms, not independence; the subsequent 'quadratic–quadratic interaction' is asserted to be of order q/m without a derivation and is referenced to missing sections S3/S6. The Stein identity (27) is then applied conditionally on D, which requires exactly the independence in question. This is the decisive step for the dominance inequality (20); it needs a complete proof.
  4. [Section 2.1, Eq. (13), definition of bσ²] The scale calibrator bσ² is defined as 'the median of {r_i^T bV^{-1} r_i} divided by its population analogue' and it is asserted to be consistent under Condition 3. The 'population analogue' is never defined, and no theorem or proof establishes consistency. Since bσ² multiplies both bΣ_URE and bΣ_RRE, it enters the risk claims in Theorems 2 and 3. Please define the quantity precisely and provide a consistency proof, or state it as an additional assumption.
minor comments (5)
  1. [Figure 7 caption] The caption says the risk reduction 'tracks the theoretical 1−q/n factor,' which contradicts Section 3.2 and Eq. (22), where the authors explicitly argue the gain is (n−d)/(n−d+q), not 1−q/n. Please correct the caption.
  2. [Proof of Theorem 2, first line] The proof says 'substitute bΣ_S = ... into (19),' but (19) is the local sequence; the loss is defined in (18). Please correct the cross-reference.
  3. [Proof of Theorem 5, reference to Theorem 2] The proof refers to 'the dominance inequality (21) of Theorem 2,' but the dominance inequality of Theorem 2 is numbered (20). This is a formatting/cross-reference error.
  4. [Table 3] The rows for 'Unrestricted, covariance (URE-cov)' and 'Restricted, covariance (RRE-cov)' report identical numbers (21.0, 39.1, 90.7) across all three tail regimes. If this is not a mistake, the table should explain why the restriction has no effect for the covariance-based estimator; otherwise, the entries should be corrected.
  5. [Algorithm 1, Step 4] The Procrustes sign correction eU ↦ eU sign diag(eU^T U) is described informally, and the claim that the alignment is 'exact in the limit' is not proved. Please provide a precise statement or a reference.

Circularity Check

1 steps flagged

Scale-calibration constant is defined via an unspecified 'population analogue'; the core spectrum and dominance results are otherwise self-contained.

specific steps
  1. other [Section 2.1, Eq. (13)]
    "bσ2 is a robust scale calibrating the trace to that of Σ (Tyler’s estimator identifies Σ only up to a positive scalar; we take bσ2 to be the median of {r⊤i bV−1ri} divided by its population analogue, which is consistent under Condition 3."

    The scale constant bσ2 is the only place scale enters in bΣRRE, and the paper's consistency claim for it rests entirely on an undefined 'population analogue.' If that phrase is read as the limiting value of the observed median, then bσ2 is consistent by definition and the claimed calibration is not an independent result. If it is read as a specific population quantity (e.g., the median of R_i^2), no definition or estimator is supplied, so the calibration step is unfalsifiable rather than derived. The spectral and risk-dominance results do not depend on this constant, so the circularity is local and does not invalidate the main distribution-free or Stein-dominance arguments.

full rationale

The central derivation chain is largely self-contained and does not reduce to its inputs. Theorem 1's distribution-free claim rests on Tyler's scale invariance, the rank-m reduction of the restricted residual projection, and the external Zhang–Cheng–Singer / Marchenko–Pastur results; no parameter is fitted to the target data and no self-citation is load-bearing. Theorem 2's dominance inequality is obtained by an explicit Stein integration-by-parts identity and a first-order expansion of the analytic shrinker in the aspect ratio; the q−2 constant is openly stated to be 'definitional' (fixed in the estimator), not data-fitted, so it is not a disguised prediction. The only definitional circularity is the scale-calibration constant bσ2 in Eq. (13), whose 'population analogue' is never specified; this is a genuine but local gap. Separately, and not a circularity, the proof of Theorem 1 Step 2 contains a likely unsupported uniform bound: the union bound over i for R_i^{-1}=O_p(1) would require n·h_n→0, while Condition 2 only gives h_n→0, and the cited supplement S1/S3/S5/S6/S8 is absent, so this step cannot be verified. That is a correctness risk, not evidence that the result is equivalent to its assumptions. Overall, the paper's main claims have independent content and rely on external theorems; the score is accordingly low.

Axiom & Free-Parameter Ledger

1 free parameters · 5 axioms · 0 invented entities

The core estimator uses no new physical or model entity; it relies on standard random-matrix theorems and the elliptical model. The main unaccounted inputs are the undefined scale-calibration constant and the assumed asymptotic independence in the Stein proof. The regularized Tyler iteration for aspect ratios above 1 is also an unstated algorithmic choice.

free parameters (1)
  • Scale calibration 'population analogue' for bσ² = not specified
    Section 2.1 defines bσ² as the median of {r_i^T V^{-1} r_i} divided by its population analogue; the analogue is never given, so any implementation must choose or estimate a constant that affects the calibrated scale of Σ̂.
axioms (5)
  • standard math Zhang et al. (2016): Tyler's M-estimator from i.i.d. spherical vectors has the same limiting spectral distribution as the Gaussian sample covariance when p/m→y∈(0,1), up to o_p(1) in operator norm.
    Invoked in the proof of Theorem 1, Step 3, to transfer the Gaussian MP law to Tyler's scatter.
  • standard math Ledoit and Wolf (2020), Theorem 3.1: analytic nonlinear shrinkage φ_c is consistent for the rotation-equivariant oracle under almost-sure weak convergence, compact support, bounded bulk density, and bandwidth n^{-1/3}.
    Used to claim that the shrunk Tyler scatter attains the oracle; appears in Theorem 1 and Theorem 3.
  • standard math Stein's lemma / Gaussian integration by parts for the conditionally Gaussian representation of elliptical errors.
    Central to the proof of Theorem 2, where the cross term is evaluated by conditioning on the scale mixture weights.
  • domain assumption Conditions 1-4: proportional regime with effective aspect ratio in (0,1), full-rank design and restriction, vanishing maximal leverage, finite second moments of radii and inverse radii, bounded spectrum of Σ.
    The whole asymptotic framework; stated in Section 3 and used throughout, especially in Theorem 1 Step 2 and Theorem 2.
  • ad hoc to paper Asymptotic independence of bΣ_r and R bB up to o(1), including negligible quadratic-quadratic coupling between the restricted residual scatter and the test statistic.
    Asserted in the proof of Theorem 2 with reference to missing 'Section S3/S6'; this independence is essential for applying Stein's identity, but the derivation is not provided in the manuscript.

pith-pipeline@v1.3.0-alltime-deepseek · 26616 in / 13797 out tokens · 142414 ms · 2026-08-01T11:34:18.204718+00:00 · methodology

0 comments
read the original abstract

We study estimation of the p*p residual scatter (shape) matrix in a high-dimensional multivariate linear regression, where p and n grow proportionally. When the coefficient matrix obeys a known linear restriction of rank q < d, as in multivariate analysis of variance, growth-curve models, and reduced-rank regression, the restricted fit leaves additional residual degrees of freedom that sharpen estimation of the shape matrix. To accommodate heavy-tailed errors, we work with independent elliptically distributed rows under a mild scale condition, a finite second moment on the radii, which is far weaker than the usual sub-Gaussian assumptions and covers every multivariate-t law with more than two degrees of freedom. Shrinking the restricted residual sample covariance directly is unsound here, since its limiting spectrum depends on the radial distribution. We instead shrink a scale-invariant scatter of the restricted residuals, whose spectrum is distribution-free over the elliptical family and obeys the same limiting law as under Gaussian errors, at a smaller effective aspect ratio. The resulting estimator attains the rotation-equivariant oracle and is asymptotically optimal within that class, and a Stein-type combination with the unrestricted estimator dominates it while remaining safe under misspecification. We further correct for the case in which the restriction is itself selected from the data. Simulations, a growth-curve experiment, and two real-data analyses illustrate the results.

Figures

Figures reproduced from arXiv: 2607.20576 by (2) Department of Statistics, Bushehr, Data Science, Faculty of Intelligent Systems Engineering, Faculty of Mathematical Sciences, Ferdowsi University of Mashhad, Hamid Karamikabir (1), Iran, Iran), Mashhad, Mohammad Arashi (2) ((1) Department of Statistics, Persian Gulf University.

Figure 1
Figure 1. Figure 1: (a) Estimation risk under a data-selected restriction: sample-splitting is costly while the full-sample correction [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Empirical spectra of the restricted residuals across Gaussian, [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (a) Ratio of restricted to unrestricted oracle risk against [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Weighted Frobenius risk against the aspect ratio [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Operator-norm error against the aspect ratio for the configuration of Figure 4. The ordering of methods is [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Relative weighted Frobenius risk against the restriction violation [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Relative weighted Frobenius risk against the restriction rank [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Bias–variance decomposition of the weighted Frobenius risk, identity target, [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Entrywise estimation error |Σbjk − Σjk| for one replication under the banded target, n = 80, p = 60. The restricted Stein estimator (right) is uniformly closer to the truth. 5 Applications We illustrate the method on a designed growth-curve analysis in which the covariate-side restriction is known and q ≤ d, exactly as the theory requires. 5.1 A designed growth-curve experiment Consider a balanced two-grou… 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

38 extracted references

  1. [1]

    ANDERSON, T. W. (2003).An Introduction to Multivariate Statistical Analysis, 3rd ed. Wiley, Hoboken

    2003

  2. [2]

    BAI, Z.ANDSILVERSTEIN, J. W. (2010).Spectral Analysis of Large Dimensional Random Matrices, 2nd ed. Springer, New York

    2010

  3. [3]

    J.ANDLEVINA, E

    BICKEL, P. J.ANDLEVINA, E. (2008). Covariance regularization by thresholding.Ann. Statist.36, 2577–2604

    2008

  4. [4]

    (2013).Concentration Inequalities: A Nonasymptotic Theory of Independence

    BOUCHERON, S., LUGOSI, G.ANDMASSART, P. (2013).Concentration Inequalities: A Nonasymptotic Theory of Independence. Oxford University Press, Oxford

    2013

  5. [5]

    BUNEA, F., SHE, Y.ANDWEGKAMP, M. H. (2011). Optimal selection of reduced rank estimators of high-dimensional matrices.Ann. Statist.39, 1282–1309

    2011

  6. [6]

    T.ANDGUO, Z

    CAI, T. T.ANDGUO, Z. (2017). Confidence intervals for high-dimensional linear regression: minimax rates and adaptivity.Ann. Statist.45, 615–646

    2017

  7. [7]

    T.ANDLIU, W

    CAI, T. T.ANDLIU, W. (2011). Adaptive thresholding for sparse covariance matrix estimation.J. Amer. Statist. Assoc. 106, 672–684

    2011

  8. [8]

    T.ANDZHOU, H

    CAI, T. T.ANDZHOU, H. H. (2012). Optimal rates of convergence for sparse covariance matrix estimation.Ann. Statist.40, 2389–2420

    2012

  9. [9]

    D., LI, B.ANDCHIAROMONTE, F

    COOK, R. D., LI, B.ANDCHIAROMONTE, F. (2010). Envelope models for parsimonious and efficient multivariate linear regression.Statist. Sinica20, 927–960

    2010

  10. [10]

    DICKER, L. H. (2014). Variance estimation in high-dimensional linear models.Biometrika101, 269–284

    2014

  11. [11]

    ELKAROUI, N. (2008). Spectrum estimation for large dimensional covariance matrices using random matrix theory. Ann. Statist.36, 2757–2790

    2008

  12. [12]

    FAN, J., GUO, S.ANDHAO, N. (2012). Variance estimation using refitted cross-validation in ultrahigh dimensional regression.J. R. Stat. Soc. Ser. B Stat. Methodol.74, 37–65

    2012

  13. [13]

    FAN, J., LIAO, Y.ANDMINCHEVA, M. (2013). Large covariance estimation by thresholding principal orthogonal complements.J. R. Stat. Soc. Ser. B Stat. Methodol.75, 603–680

    2013

  14. [14]

    FAN, J., LIU, H.ANDWANG, W. (2018). Large covariance estimation through elliptical factor models.Annals of Statistics,46(4), 1383–1414

    2018

  15. [15]

    IZENMAN, A. J. (1975). Reduced-rank regression for the multivariate linear model.J. Multivariate Anal.5, 248–264

    1975

  16. [16]

    G.ANDBOCK, M

    JUDGE, G. G.ANDBOCK, M. E. (1978).The Statistical Implications of Pre-Test and Stein-Rule Estimators in Econometrics. North-Holland, Amsterdam

    1978

  17. [17]

    KOLTCHINSKII, V.ANDLOUNICI, K. (2017). Concentration inequalities and moment bounds for sample covariance operators.Bernoulli23, 110–133. 23 Restricted nonlinear shrinkage of covarianceA PREPRINT

    2017

  18. [18]

    LEDOIT, O.ANDWOLF, M. (2004). A well-conditioned estimator for large-dimensional covariance matrices.J. Multivariate Anal.88, 365–411

    2004

  19. [19]

    LEDOIT, O.ANDWOLF, M. (2012). Nonlinear shrinkage estimation of large-dimensional covariance matrices.Ann. Statist.40, 1024–1060

    2012

  20. [20]

    LEDOIT, O.ANDWOLF, M. (2020). Analytical nonlinear shrinkage of large-dimensional covariance matrices.Ann. Statist.48, 3043–3065. MAR ˇCENKO, V. A.ANDPASTUR, L. A. (1967). Distribution of eigenvalues for some sets of random matrices.Math. USSR-Sb.1, 457–483

    2020

  21. [21]

    REID, S., TIBSHIRANI, R.ANDFRIEDMAN, J. (2016). A study of error variance estimation in lasso regression.Statist. Sinica26, 35–67

    2016

  22. [22]

    RUDELSON, M.ANDVERSHYNIN, R. (2013). Hanson–Wright inequality and sub-Gaussian concentration.Electron. Commun. Probab.18, no. 82, 1–9

    2013

  23. [23]

    SALEH, A. K. M. E. (2006).Theory of Preliminary Test and Stein-Type Estimation with Applications. Wiley, Hoboken

    2006

  24. [24]

    W.ANDBAI, Z

    SILVERSTEIN, J. W.ANDBAI, Z. D. (1995). On the empirical distribution of eigenvalues of a class of large dimensional random matrices.J. Multivariate Anal.54, 175–192

    1995

  25. [25]

    W.ANDCHOI, S

    SILVERSTEIN, J. W.ANDCHOI, S. I. (1995). Analysis of the limiting spectral distribution of large dimensional random matrices.J. Multivariate Anal.54, 295–309

    1995

  26. [26]

    STEIN, C. M. (1981). Estimation of the mean of a multivariate normal distribution.Ann. Statist.9, 1135–1151

    1981

  27. [27]

    SUN, T.ANDZHANG, C.-H. (2012). Scaled sparse linear regression.Biometrika99, 879–898

    2012

  28. [28]

    RAMEY, J. A. (2016).datamicroarray: Collection of Data Sets for Classification. R package and data collection; https://github.com/ramey/datamicroarray

    2016

  29. [29]

    TSYBAKOV, A. B. (2009).Introduction to Nonparametric Estimation. Springer, New York

    2009

  30. [30]

    DUA, D.ANDGRAFF, C. (2019). UCI Machine Learning Repository: Communities and Crime Data Set. University of

    2019

  31. [31]

    (2018).High-Dimensional Probability: An Introduction with Applications in Data Science

    VERSHYNIN, R. (2018).High-Dimensional Probability: An Introduction with Applications in Data Science. Cambridge University Press, Cambridge

    2018

  32. [32]

    WANG, W.ANDFAN, J. (2017). Asymptotics of empirical eigenstructure for high dimensional spiked covariance.Ann. Statist.45, 1342–1374

    2017

  33. [33]

    (2015).Large Sample Covariance Matrices and High-Dimensional Data Analysis

    YAO, J., ZHENG, S.ANDBAI, Z. (2015).Large Sample Covariance Matrices and High-Dimensional Data Analysis. Cambridge University Press, Cambridge

    2015

  34. [34]

    E., NAEVE, C., WONG, L.ANDDOWNING, J

    LIU, H., PUI, C.-H., EVANS, W. E., NAEVE, C., WONG, L.ANDDOWNING, J. R. (2002). Classification, subtype discovery, and prediction of outcome in pediatric acute lymphoblastic leukemia by gene expression profiling.Cancer Cell1, 133–143

    2002

  35. [35]

    YU, G.ANDBIEN, J. (2019). Estimating the error variance in a high-dimensional linear model.Biometrika106, 533–546

    2019

  36. [36]

    TYLER, D. E. (1987). A distribution-freeM-estimator of multivariate scatter.Ann. Statist.,15(1), 234–251

    1987

  37. [37]

    ZHANG, T., CHENG, X.ANDSINGER, A. (2016). Marchenko–Pastur law for Tyler’s M-estimator.J. Multivariate Anal.,149, 114–123

    2016

  38. [38]

    REDMOND, M.ANDBAVEJA, A. (2002). A data-driven software tool for enabling cooperative information sharing among police departments.European Journal of Operational Research,141, 660–678. 24

    2002