Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I would like to prove the following theorem:

Let $H\in (0,1)$. The fractional Brownian motion $B_H$ admits a version whose sample paths are $a.s.$ Hölder continuous of order strict less than $H$.

For any $\alpha >0$ we have by self similarity $$\mathbb{E}[|B_H(t)-B_H(s)|^{\alpha}]=\mathbb{E}[|B_H(1)|^{\alpha}]|t-s|^{\alpha H}.$$

The proof is done after applying the criterion of Kolmogorov which says:

A process $(X_t)_{t\in\mathbb{R}}$ admits a continuous modification if there exist constants $a,b,k>0$ such that $$\mathbb{E}[|X(t)-X(s)|^a]\leq k|t-s|^{1+b}$$ for all $s,t\in\mathbb{R}$.

But I don't know how to apply this criterion. Any help please.

Edit:

Well, maybe I should try to state my problem more precisely. I just would like to understand the proof. If I use the criterion of Kolmogorov it should hold $$\mathbb{E}[|B_H(t)-B_H(s)|^{\alpha}]=\mathbb{E}[|B_H(1)|^{\alpha}]|t-s|^{\alpha H}\leq k|t-s|^{1+\beta}$$ for $\alpha,\beta,k>0$, right? I don't see any relationship to the Hölder continuity.

Is there nobody who can demonstrate the proof for me to understand?

Maybe I should set $k=\mathbb{E}[|B_H(1)|^{\alpha}]$ and $\beta=\alpha H$ and say that $B_H$ is $\gamma$-Hölder continuous for every $\gamma\in\big[0,{\alpha H\over \alpha}\big)$?

share|improve this question
    
In fact, there Kolmogorov criterion in your formulation can not help apriori. Use this one galton.uchicago.edu/~lalley/Courses/385/GaussianProcesses.pdf on the page six –  Tarasenya Dec 17 '12 at 22:17
2  
or for instance, here statslab.cam.ac.uk/~beresty/teach/StoCal/sc3.pdf page 10. Good lecture notes, by the way. –  Tarasenya Dec 17 '12 at 22:22

1 Answer 1

There exists an extension of Kolmogorov's continuity criterion.

Theorem 1.4. of these lecture notes by Nathanaël Berestycki states the following:

Theorem. Let $(X_t)_{t\in[0,1]}$ be a real-valued stochastic process. Suppose that there exists positive constants $p,c,\varepsilon$ such that for each $s,t\in[0,1]$, $$\mathbb E\left[\left|X_t-X_s\right|^p\right]\leqslant c\left|t-s\right|^{1+\varepsilon}.$$ Then there exists a modification $\widetilde X$ of $X$ which is almost surely $\alpha$-Hölder continuous for any $\alpha\in (0,\varepsilon /p)$.

When $X=B_H$, we have
$$\mathbb E\left[\left|X_t-X_s\right|^p\right]=\left|t-s\right|^{pH}\mathbb E\left[\left|B_H(1)\right|^p\right],$$ hence we can use the theorem with $p:=(1+R)/H$ (for a fixed $R$) $c:=\mathbb E\left[\left|B_H(1)\right|^p\right]$, $\varepsilon:=pH-1=R $, which provides an $\alpha$-Hölder continuous modification for each $$\alpha\lt \frac{\varepsilon}p= H\frac{R}{1+R} .$$ As $R$ is arbitrary, we can get a modification for any exponent strictly smaller than $H$.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.