Continuity of some anticipating integral processes (Q1379912)

From MaRDI portal
Revision as of 21:57, 19 July 2023 by Importer (talk | contribs) (‎Created a new Item)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
scientific article
Language Label Description Also known as
English
Continuity of some anticipating integral processes
scientific article

    Statements

    Continuity of some anticipating integral processes (English)
    0 references
    0 references
    0 references
    24 June 1998
    0 references
    Let \(u\) be a stochastic process in the Sobolev space \(\mathbb{L}^{1,2}\). Then, the Skorokhod anticipating integral \(X_t=\int_0^t u_\theta dW_\theta\) exists for every \(0\leq t\leq 1\), and the question arises whether the process \(X_t\) has a continuous version. This paper establishes two criteria for an affirmative answer: 1) \(u\in\mathbb{L}^{2,2}\) and \(\int_0^1\text{E}| u_\theta|^\beta d\theta<\infty\), for some \(\beta>2\). The proof is based on Garsia-Rodemish-Rumsey lemma and a particular case of the following \(L^p\)-estimate for \(X_t\), which may have independent interest: If \(2\leq p<\infty\), \(p/2\leq q\), \(1/q+1/r=2/p\), and \(u_\theta\in\mathbb{L}^{2,2}\) satisfies \[ \int_0^1\text{E}| u_\theta|^{p\vee r} d\theta<\infty \quad\text{and}\quad \int_0^1 \text{E}\Big\{\int_0^1\int_0^1 | D_sD_tu_\theta|^2 d\theta dt\Big\}^{q/2} ds<\infty, \] then \(\int_a^b u_{\theta} dW_\theta\) belongs to \(L^p\) for all \(0\leq a<b\leq 1\) and \[ \begin{aligned} \text{E}\Big|\int_a^b u_{\theta} dW_{\theta}\Big|^p &\leq c_p(b-a)^{p/2-1}\Big\{\int_a^b \text{E}| u_{\theta}|^p d\theta+ \int_a^b \text{E}| u_{\theta}|^r d\theta \\ &+ M_q\int_0^1\text{E}\Big(\int_a^b| D_su_{\theta}|^2 d\theta\Big)^{q/2} ds\\ &+ M_q\int_0^1\text{E}\Big(\int_a^b\int_0^1 | D_tD_su_{\theta}|^2 dt d\theta\Big)^{q/2} ds \Big\} \;, \end{aligned} \] for some constants \(c_p\) and \(M_q\). 2) \(u\in\mathbb{L}^{1,2}\), \(\int_0^1 \text{E}| u_{\theta}|^{\beta} d\theta<\infty\) for some \(\beta>2\), and \(\int_0^1\int_0^1 \text{E}| D_ru_{\theta}|^2 (r\lor\theta-r\land\theta)^{-\varepsilon} dr d\theta<\infty\), for some \(\varepsilon>0\). The proof uses Clark--Ocone formula and the fact that a process \(\{X_t,\;t\in[0,1]\}\) such that \(\sum_{n=1}^{\infty}\sum_{k=0}^{q^n-1} P\big(| X_{(k+1)q^{-n}}-X_{kq^{-n}}| \geq q^{-\gamma n}\big)<\infty\) for some \(\gamma>0\) and some integer \(q>1\) has a continuous version. Other results in the same direction can be found in [\textit{D. Nualart}, ``The Malliavin calculus and related topics'' (1995; Zbl 0837.60050)]. The question whether the sole hypothesis \(u\in\mathbb{L}^{1,2}\) implies that the indefinite integral process \(X_t\) has a continuous version remains open.
    0 references
    0 references
    Skorokhod integral
    0 references
    Sobolev space
    0 references
    Clark-Ocone formula
    0 references
    Malliavin calculus
    0 references