Annales de l'Institut Henri Poincaré, Probabilités et Statistiques

Comparison between two types of large sample covariance matrices

Guangming Pan

Full-text: Open access


Let $\{X_{ij}\}$, $i,j=\cdots$, be a double array of independent and identically distributed (i.i.d.) real random variables with $EX_{11}=\mu$, $E|X_{11}-\mu|^{2}=1$ and $E|X_{11}|^{4}<\infty$. Consider sample covariance matrices (with/without empirical centering) $\mathcal{S}=\frac{1}{n}\sum_{j=1}^{n}(\mathbf{s}_{j}-\bar{\mathbf{s}})(\mathbf{s}_{j}-\bar{\mathbf{s}})^{T}$ and $\mathbf{S} =\frac{1}{n}\sum_{j=1}^{n}\mathbf{s}_{j}\mathbf{s}_{j}^{T}$, where $\bar{\mathbf{s}}=\frac{1}{n}\sum_{j=1}^{n}\mathbf{s}_{j}$ and $\mathbf{s}_{j}=\mathbf{T} _{n}^{1/2}(X_{1j},\ldots,X_{pj})^{T}$ with $(\mathbf{T} _{n}^{1/2})^{2}=\mathbf{T} _{n}$, non-random symmetric non-negative definite matrix. It is proved that central limit theorems of eigenvalue statistics of $\mathcal{S}$ and $\mathbf{S} $ are different as $n\rightarrow\infty$ with $p/n$ approaching a positive constant. Moreover, it is also proved that such a different behavior is not observed in the average behavior of eigenvectors.


Soit $\{X_{ij}\}$, $i,j=1,2,\ldots$, un tableau à double entrées, les $X_{ij}$ étant des variables aléatoires réelles indépendantes et identiquement distribuées (i.i.d.) et où $\mathbf{E} X_{11}=\mu$, $\mathbf{E} \vert X_{11}-\mu\vert ^{2}=1$ et $\mathbf{E} |X_{11}|^{4}<\infty$. Considérons les matrices de covariances empiriques suivantes (avec/sans centrage empirique): $\mathcal{S}=\frac{1}{n}\sum^{n}_{j=1}(\mathbf{s} _{j}-\bar{ \mathbf {s}})(\mathbf{s} _{j}-\bar{ \mathbf {s}})^{T}$ et $\mathbf{S}=\frac{1}{n}\sum^{n}_{j=1}\mathbf{s} _{j}\mathbf{s} _{j}^{T}$, avec $\bar{ \mathbf {s}}=\frac{1}{n}\sum^{n}_{j=1}\mathbf{s} _{j}$ et $\mathbf{s} _{j}=\mathbf{T}^{1/2}_{n}(X_{1j},\ldots,X_{pj})^{T}$, où $(\mathbf{T}^{1/2}_{n})^{2}=\mathbf{T}_{n}$ est une matrice déterministe définie positive. Nous démontrons que, sous le régime asymptotique $n\rightarrow\infty$ et $p/n$ converge vers une constante positive, le théorème central limite pour la statistique $\mathcal{S}$ est différent de celui concernant la statistique $\mathbf{S}$. En outre, nous montrons que cette différence de comportement n’est pas observée pour le comportement moyen des vecteurs propres.

Article information

Ann. Inst. H. Poincaré Probab. Statist., Volume 50, Number 2 (2014), 655-677.

First available in Project Euclid: 26 March 2014

Permanent link to this document

Digital Object Identifier

Mathematical Reviews number (MathSciNet)

Zentralblatt MATH identifier

Primary: 15A52 60F15: Strong theorems 62E20: Asymptotic distribution theory 60F17: Functional limit theorems; invariance principles

Central limit theorems Eigenvectors and eigenvalues Sample covariance matrix Stieltjes transform Strong convergence


Pan, Guangming. Comparison between two types of large sample covariance matrices. Ann. Inst. H. Poincaré Probab. Statist. 50 (2014), no. 2, 655--677. doi:10.1214/12-AIHP506.

Export citation


  • [1] T. W. Anderson. An Introduction to Multivariate Statistical Analysis, 2nd edition. Wiley, New York, 1984.
  • [2] Z. D. Bai and J. W. Silverstein. No eigenvalues outside the support of the limiting spectral distribution of large dimensional random matrices. Ann. Probab. 26 (1998) 316–345.
  • [3] Z. D. Bai and J. W. Silverstein. CLT for linear spectral statistics of large dimensional sample covariance matrices. Ann. Probab. 32 (2004) 553–605.
  • [4] Z. D. Bai, B. Q. Miao and G. M. Pan. On asymptotics of eigenvectors of large sample covariance matrix. Ann. Probab. 35 (2007) 1532–1572.
  • [5] Z. D. Bai and J. W. Silverstein. Spectral Analysis of Large Dimensional Random Matrices, 2nd edition. Springer, New York, 2010.
  • [6] Z. D. Bai and G. M. Pan. Limiting behavior of eigenvectors of large Wigner matrices. J. Stat. Phys. 146 (2012) 519–549.
  • [7] F. Benaych-Georges. Eigenvectors of Wigner matrices: Universality of global fluctuations. Available at
  • [8] P. Billingsley. Convergence of Probability Measures. Wiley, New York, 1968.
  • [9] C. Donati-Martin and A. Rouault. Truncations of Haar unitary matrices, traces and bivariate Brownian bridge. RMTA 1(1) (2012) 1150007.
  • [10] I. Johnstone. On the distribution of the largest eigenvalue in principal component analysis. Ann. Statist. 29 (2001) 295–327.
  • [11] T. F. Jiang. The limiting distriubtions of eigenvalues of sample correlation matrices. Sankhya 66 (2004) 35–48.
  • [12] D. Jonsson. Some limit theorems of the eigenvalues of sample convariance matrix. J. Multivariate Anal. 12 (1982) 1–38.
  • [13] O. Ledoit and S. Peche. Eigenvectors of some large sample covariance matrix ensembles. Probab. Theory Related Fields 151 (2011) 233–264.
  • [14] V. A. Marčenko and L. A. Pastur. Distribution for some sets of random matrices. Math. USSR-Sb. 1 (1967) 457–483.
  • [15] G. M. Pan and W. Zhou. Central limit theorem for Hotelling’s $T^{2}$ statistic under large dimension. Ann. Appl. Probab. 21 (2011) 1860–1910.
  • [16] G. M. Pan and W. Zhou. Central limit theorem for signal-to-interference ratio of reduced rank linear receiver. Ann. Appl. Probab. 18 (2008) 1232–1270.
  • [17] G. M. Pan, Q. M. Shao and W. Zhou. Universality of sample covariance matrices: CLT of the smoothed empirical spectral distribution. Preprint.
  • [18] J. W. Silverstein. On the eigenvectors of large dimensional sample covariance matrices. J. Multivariate Anal. 30 (1989) 1–16.
  • [19] J. W. Silverstein. Strong convergence of the limiting distribution of the eigenvalues of large dimensional random matrices. J. Multivariate Anal. 55 (1995) 331–339.
  • [20] J. W. Silverstein. Weak convergence of random functions defined by the eigenvectors of sample covariance matrices. Ann. Probab. 18 (1990) 1174–1194.
  • [21] T. Tao and V. Vu. Random matrices: Universal properties of eigenvectors. Available at arXiv:1103.2801v2.
  • [22] H. Xiao and W. Zhou. On the limit of the smallest eigenvalue of some sample covariance matrix. J. Theoret. Probab. 23 (2010) 1–20.
  • [23] Y. Q. Yin, Z. D. Bai and P. R. Krishanaiah. On the limit of the largest eigenvalue of the large dimensional sample covariance matrix. Probab. Theory Related Fields 78 (1988) 509–521.