2021/06/01 by Marcus, Michael B., Rosen, Jay
#60F15 #60G15 #60G17 #FOS: Mathematics #Probability (math.PR)
paper · doi:10.48550/arxiv.2106.00542
Let η=\η(t);t∈ [0,1]\ be a mean zero continuous Gaussian process with covariance U=\U(s,t),s,t∈ [ 0,1]\, with U(0,0)>0. Let \ηi;i=1,…, k\ be independent copies of η and set Yk(t)=∑i=1k η2i(t), t∈ [ 0,1]. The stochastic process Yk =\Yk (t),t∈ [ 0,1] \ is referred to as a chi--square process of order k with kernel U. Let ϕ(t) be a positive function on [0,δ] for some δ>0. If \limsupt→ 0( η(t)-η(0))/( ϕ(t) )=1 a.s., then for all integers k≥ 1, \limsupt→ 0 \fracYk (t)-Yk (0) ϕ(t) = 2 Y1/2k(0) a.s. Set σ2(u,v)=E(η(u)-η(v))2\quadand \widetildeσ2(x)=sup|u-v|≤ xσ2(u,v). Assume that inft∈ [0,1]U(t,t)>0 and, limx→0\widetildeσ2(x)log 1/x =0. Let φ(t) be a positive function on [0,1]. Then if limh→ 0sup_\stackrel|u-v|≤ h u,v∈Δ( η(u)-η(v))/( φ(|u-v|) )=1 a.s. for all intervals Δ⊂ [0,1], it follows that for all intervals Δ⊂ [0,1] and all integers k≥ 1, limh→ 0sup_\stackrel|u-v|≤ h u,v∈Δ \fracYk (u)-Yk (v) φ(|u-v|) = 2 supu∈ΔYk 1/2(u), \hspace.2 ina.s.