Simple setting for white noise calculus using Bargmann space and Gauss transform (Q1894028): Difference between revisions

Let \(E_0\) be a real separable Hilbert space with \(\dim E_0=\infty\) and norm \(\|\cdot,\cdot\|_0\). Suppose that \(D>1\) is a densely defined selfadjoint operator on \(E_0\), whose inverse is of Hilbert-Schmidt type. Let \(H\subset H_0\subset H^*\) be the complexification of the real Gel'fand triplet \(E\subset E_0\subset E^*\) of \(E_0\) that is generated by the norms \(\|f\|=\|D^pf\|_0\), \(p\in\mathbb{R}\). Consider the measure \(\mu\) determined by \(\int_{E^*}\exp[i\langle x,\xi\rangle] d\mu=\exp[-(1/2)\|\xi\|^2_0]\). Then \(L^2(E^*,\mu)\) is the space of white noise functionals. Denote by \({\mathcal P}(H^*)\) the space of polynomials in \(\langle x,\xi\rangle\), \(x\in E^*\), \(\xi\in H\), and let \(G:{\mathcal P}(E^*)\to{\mathcal P}(H^*)\) be the map \(G\varphi(w)=\int_{E^*}\varphi(x+w/\sqrt2) d\mu(x)\). Consider the system of operators \(\{G^{-1}\Lambda(D^p)G\), \(p\in\mathbb{R}\}\), where \(\Lambda(D^p)f(z)=f(D^pz)\) for \(f\in{\mathcal P}(H^*)\), \(z\in H^*\), \(p\in\mathbb{R}\). The author shows that the completions of \({\mathcal P}(E^*)\) under the norms \(\|f\|^2_{(L^2)_p}=\| G^{-1}\Lambda(D^p)Gf\|^2_{(L^2)_0}= \int_{E^*}|(G^{-1}D^pGf)(z))|^2 d\mu(z)\) generates a Gel'fand triplet \(({\mathcal S})\subset(L^2)\subset({\mathcal S}')\) of \((L^2)\), that can provide a simplified approach to white noise functionals. The triple is used to establish a delicate characterization of positive white noise functionals, based on a Fernique type estimate for their associated measures.