English

On multiplier processes under weak moment assumptions

Statistics Theory 2016-01-26 v1 Statistics Theory

Abstract

We show that if VRnV \subset \R^n satisfies a certain symmetry condition (closely related to unconditionaity) and if XX is an isotropic random vector for which \inrX,tLpLp\|\inr{X,t}\|_{L_p} \leq L \sqrt{p} for every tSn1t \in S^{n-1} and plognp \lesssim \log n, then the corresponding empirical and multiplier processes indexed by VV behave as if XX were LL-subgaussian.

Cite

@article{arxiv.1601.06523,
  title  = {On multiplier processes under weak moment assumptions},
  author = {Shahar Mendelson},
  journal= {arXiv preprint arXiv:1601.06523},
  year   = {2016}
}
R2 v1 2026-06-22T12:35:53.114Z