English

The growth of additive processes

Probability 2011-11-10 v1

Abstract

Let XtX_t be any additive process in Rd.\mathbb{R}^d. There are finite indices δi,βi,i=1,2\delta_i, \beta_i, i=1,2 and a function uu, all of which are defined in terms of the characteristics of XtX_t, such that \liminf_{t\to0}u(t)^{-1/\eta}X_t^*= \cases{0, \quad if η>δ1\eta>\delta_1, \cr\infty, \quad if η<δ2\eta<\delta_2,} \limsup_{t\to0}u(t)^{-1/\eta}X_t^*= \cases{0, \quad if η>β2\eta>\beta_2, \cr\infty, \quad if η<β1\eta<\beta_1,}\qquad {a.s.}, where Xt=sup0stXs.X_t^*=\sup_{0\le s\le t}|X_s|. When XtX_t is a L\'{e}vy process with X0=0X_0=0, δ1=δ2\delta_1=\delta_2, β1=β2\beta_1=\beta_2 and u(t)=t.u(t)=t. This is a special case obtained by Pruitt. When XtX_t is not a L\'{e}vy process, its characteristics are complicated functions of tt. However, there are interesting conditions under which uu becomes sharp to achieve δ1=δ2\delta_1=\delta_2, β1=β2.\beta_1=\beta_2.

Keywords

Cite

@article{arxiv.0707.3886,
  title  = {The growth of additive processes},
  author = {Ming Yang},
  journal= {arXiv preprint arXiv:0707.3886},
  year   = {2011}
}

Comments

Published at http://dx.doi.org/10.1214/009117906000000593 in the Annals of Probability (http://www.imstat.org/aop/) by the Institute of Mathematical Statistics (http://www.imstat.org)

R2 v1 2026-06-21T09:01:59.259Z