On the principal components of sample covariance matrices

We introduce a class of $$M \times M$$M×M sample covariance matrices $${\mathcal {Q}}$$Q which subsumes and generalizes several previous models. The associated population covariance matrix $$\Sigma = \mathbb {E}{\mathcal {Q}}$$Σ=EQ is assumed to differ from the identity by a matrix of bounded rank. All quantities except the rank of $$\Sigma - I_M$$Σ-IM may depend on $$M$$M in an arbitrary fashion. We investigate the principal components, i.e. the top eigenvalues and eigenvectors, of $${\mathcal {Q}}$$Q. We derive precise large deviation estimates on the generalized components $$\langle {\mathbf{{w}}} , {\varvec{\xi }_i}\rangle $$⟨w,ξi⟩ of the outlier and non-outlier eigenvectors $$\varvec{\xi }_i$$ξi. Our results also hold near the so-called BBP transition, where outliers are created or annihilated, and for degenerate or near-degenerate outliers. We believe the obtained rates of convergence to be optimal. In addition, we derive the asymptotic distribution of the generalized components of the non-outlier eigenvectors. A novel observation arising from our results is that, unlike the eigenvalues, the eigenvectors of the principal components contain information about the subcritical spikes of $$\Sigma $$Σ. The proofs use several results on the eigenvalues and eigenvectors of the uncorrelated matrix $${\mathcal {Q}}$$Q, satisfying $$\mathbb {E}{\mathcal {Q}} = I_M$$EQ=IM, as input: the isotropic local Marchenko–Pastur law established in Bloemendal et al. (Electron J Probab 19:1–53, 2014), level repulsion, and quantum unique ergodicity of the eigenvectors. The latter is a special case of a new universality result for the joint eigenvalue–eigenvector distribution.