Quantitative spectral rate for the normal-scores covariance

Let S^n\widehat{\bm{S}}_n be the normal-scores covariance, and let mnm_n and mFm_F denote its empirical Stieltjes transform and the limiting generalized Marčenko--Pastur Stieltjes transform, respectively. Assume --. Quantitative spectral rate. There is a constant C>0C>0 such that, for every fixed η0>0\eta_0>0,

sup⁡ℑz≥η0∣mn(z)−mF(z)∣=OP(n−1/2),\sup_{\Im z\ge\eta_0}|m_n(z)-m_F(z)|=O_{\mathbb{P}}(n^{-1/2}),

and, more generally, a local Marčenko--Pastur law holds for S^n\widehat{\bm{S}}_n down to spectral scales ℑz≳n−1+ε\Im z\gtrsim n^{-1+\varepsilon}. Establishing this would require an anisotropic/local-law analysis tracking the rank-perturbation error inside the resolvent; this rate is not established by the results stated in the paper.

References

Primary source

Hamid Karamikabir and Mohammad Arashi, “Mens: Nonlinear shrinkage estimation in nonparanormal models for financial applications”, arXiv:2607.19825 (2026).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.