Answers are grounded in this paper and cite their sources.
Consider a Gaussian vector z=(x',y')', consisting of two sub-vectors x and y with dimensions p and q, respectively. With n independent observations of z, we study the correlation between x and y, from the perspective of the canonical correlation analysis. We investigate the high-dimensional case: both p and q are proportional to the sample size n. Denote by Σᵤᵥ the population cross-covariance matrix of random vectors u and v, and denote by Sᵤᵥ the sample counterpart. The canonical correlation coefficients between x and y are known as the square roots of the nonzero eigenvalues of the canonical correlation matrix Σₓₓ⁻¹ΣxyΣyy⁻¹Σyx. In this paper, we focus on the case that Σxy is of finite rank k, that is, there are k nonzero canonical correlation coefficients, whose squares are denoted by r₁≥⋯≥ rₖ>0. We study the sample counterparts of rᵢ,i=1,…,k, that is, the largest k eigenvalues of the sample canonical correlation matrix Sₓₓ⁻¹SxySyy⁻¹Syx, denoted by λ₁≥⋯≥λₖ. We show that there exists a threshold rc∈(0,1), such that for each i∈\1,…,k\, when rᵢ≤ rc, λᵢ converges almost surely to the right edge of the limiting spectral distribution of the sample canonical correlation matrix, denoted by d₊. When rᵢ>rc, λᵢ possesses an almost sure limit in (d₊,1], from which we can recover rᵢ’s in turn, thus provide an estimate of the latter in the high-dimensional scenario. We also obtain the limiting distribution of λᵢ’s under appropriate normalization. Specifically, λᵢ possesses Gaussian type fluctuation if rᵢ>rc, and follows Tracy–Widom distribution if rᵢ<rc. Some applications of our results are also discussed.
No takes yet. Share an insight, caveat, or question.
Bao et al. (2018) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: