Convergence in variation of the joint laws of multiple Wiener-Itô integrals (Q850203)

Let \(I_d(f)=\int_{T^d}f(t_1,\dots,t_d)\,dW(t_1)\dots dW(t_d)\) be the multiple Wiener-Itô integral over some measure space \((T,{\mathcal T},\tau)\), \(f\in H_d:= L^2(T^d,{\mathcal T}^{\otimes d},\tau^{\otimes d})\). Assume that a sequence \(((f_1^n,\dots,f_p^n))_{n\in{\mathbb N}}\) of elements of the Hilbert space \(H_{d_1}\times\dots\times H_{d_p}\) converges to \((f_1,\dots,f_p)\). A condition (rather complicated) is given under which the laws of \((I_{d_1}(f_1^n),\dots,I_{d_p}(f_p^n))\) converge in variation to the law of \((I_{d_1}(f_1),\dots,I_{d_p}(f_p))\). This condition in particular implies that the law of \((I_{d_1}(f_1),\dots,I_{d_p}(f_p))\) has density w.r.t. the Lebesgue measure in \({\mathbb R}^p\). In the case \(p=1\) this condition reduces to \(f\neq 0\) a.e., recovering the convergence part of a theorem by \textit{Yu. A. Davydov} and \textit{G. V. Martynova} [in: Statistics and control of stochastic processes, Collect. Artic., Moskva, 55--57 (1989; Zbl 0709.60056)]. However, the proof of the present, more general, result is different, based on the so called superstructure method, developed by \textit{Yu. A. Davydov, M. A. Lifshits} and \textit{N. V. Smorodina} [``Local properties of distributions of stochastic functionals'' (1998; Zbl 0897.60042)].