Multiple integrals with respect to \(L\)-mixing processes (Q1210288)

Let \((x_ t)\) be an \(R^ k\)-valued stochastic process. We consider a family \((F_ t)\), \(t\geq 0\), of monotone increasing \(\sigma\)-algebras and a family \((F^ +_ t)\), \(t\geq 0\), of monotone decreasing \(\sigma\)- algebras such that \(F_ t\), \(F^ +_ t\) are independent for all \(t\) and \(F^ +_ t=\sigma\bigl(\bigcup_{\varepsilon>0}F^ +_{t+\varepsilon}\bigr)\) for all \(t\). We say that \((x_ t)\) is \(L\)- mixing with respect to \((F_ t,F^ +_ t)\) if it is \((F_ t)\)- progressively measurable and for \(q\) \((\geq 1)\) \[ \sup_{t\geq 0}E^{1/q}| x_ t|^ q<\infty\quad\text{and }\int^ \infty_ 0\sup_{t\geq\tau}E^{1/q}| x_ t-E(x_ t| F^ t_{t- \tau})|^ qd\tau<\infty. \] The concept of \(L\)-mixing process has been developped in the statistical theory of linear stochastic systems. The author obtains a moment inequality for multiple integrals of the form \[ \int^ T_ 0\int^{t_ N}_ 0\cdots\int^{t_ 2}_ 0(f_{1,t_ 1}u_{1,t_ 1})\cdots(f_{N,t_ N} u_{N,t_ N})dt_ 1\cdots dt_ N, \] where \((u_{k,t})\) are zero-mean \(L\)-mixing processes and \((f_{k,t})\) are locally square-integrable deterministic functions. Examples are shown.