Iterated integrals with respect to Bessel processes (Q2387335)

The authors work with a filtered complete probability \((\Omega,\mathcal{F},\mathcal{F_t},P)\) satisfying the usual conditions, and let \(B=(B_t)_{t \geq 0}\) be a standard Brownian motion starting at zero. For any continuous process \(X\) we denote \(X_t^*=\sup_{0 \leq s \leq t} | X_s| \) and \(X*=X_{\infty}^*\). The inequalities \(c_{p,\delta} \| \tau\| p \leq \| X_{\tau}^*\| p \leq C_{p,\delta} \| \tau\| p\) and extension of inequalities are first considered by \textit{S. E. Graversen} and \textit{G. Peškir} [J. Inequal. Appl. 2, 99--119 (1998; Zbl 0903.60074) and Proc. Am. Math. Soc. 128, 3035--3041 (2000; Zbl 0954.60062)]. In this paper, the authors consider inequalities for iterated integrals: \( I_n (t,\delta)= \int_0^t I_{n-1} (s, \delta) dX_s.\) The authors prove the following theorems: Theorem 1.1. Let \(X=(X_t)_{t \geq 0}\) be the square of a \(\delta\)-dimensional Bessel process starting at zero. Define iterated stochastic integrals \(I_n(t,\delta), t\geq 0\), inductively by \(I_n(t,\delta)=\int_0^t I_{n-1}(s,\delta) dX_s\) with \(I_0(t,\delta)=1\) and \(I_1(t,\delta)=X_t.\) Then the inequalities \[ c_{m,p,\delta} \| \tau^n\|_p \leq \biggl\| \sup_{0 \leq t \leq \tau} I_n(t,\delta)\biggr\| _{p} \leq C_{n,p,\delta} \| \tau^n\| _p \] hold for all \(0 <p < +\infty\) and all stopping times \(\tau.\) Theorem 1.2. Let \(G_{\delta} (t)=\log(1+\delta \log(1+t)), \delta >0.\) Under the condition of Theorem 1.1., we have \[ c_{m,p,\delta} \| G_{\delta}(\tau)^n\| _{p} \leq \biggl\| \sup_{0 \leq t \leq \tau} \frac{| I_n(t,\delta)| }{(1+t)^n}\biggr\| _{p} \leq C_{n,p,\delta} \| G_{\delta}(\tau)^n\| _{p} \] for all \(0< p <\infty\) and all stopping times \(\tau.\)