Integral transforms of functionals in \(L^{2}(C_{a,b}[0,T])\) (Q1040680): Difference between revisions

A generalized Brownian motion \(Y\) on \([0,T]\) is a Gaussian process, the finite dimensional marginal distributions of which are normal laws with shifts resp. variances defined by real functions \(a, b\), where \(a\) is absolutely continuous with \(a'\in L^2[0,T]\) and \(b\) strictly increasing with positive continuous derivative \(b'\). Let \(\mu\) denote the corresponding measure on the path space, w.l.o.g. concentrated on a function space \(C_{a,b}[0,T] \cong \{x\in C[0,T] : x(0)=0\}\). Let \(L_{a,b}^2[0,t]\) denote the \(L^2\)-space w.r.t. the measure defined by \(|a|+b\). In analogy to the classical setup, a (generalized) Paley-Wiener-Zygmund stochastic integral \(\langle v,x\rangle\) is defined for \(v\in L^2_{a,b}[0,T]\) and \(\mu\)-a.a. \(x\in C_{a,b}[0,T]\) by an orthonormal Hilbert-space basis of \(L_{a,b}^2[0,T]\) [see, \textit{S. J. Chang} and \textit{D. L. Skoug}, Integral Transforms Spec. Funct. 14, No.~5, 375--393 (2003; Zbl 1043.28014)]. For Borel measurable functions \(F\) on \(C_{a,b}[0,T]\) (called as usual \textit{functionals}) expectation is defined as \(E(F):= \int F(x) \,d \mu(x)\). For a given ONS \(\{\alpha_n\}\) in \(L^2(\mathbb{R})\) the authors investigate generalized Hermitian polynomials \(H_{m,j}\) (adapted to \(a\) and \(b\)) and Hermitian functions \(K_{m,j}\), defining a complete ONS in \(L^2(\mathbb{R})\) (w.r.t. Lebesgue measure), and finally generalized Fourier-Hermitian functionals \(\varphi_{(m,k)}:= H_{m,k}(\langle \alpha_k, \cdot\rangle )\) and finite products \(\varphi_{m_1, \dots m_k} := \prod \varphi_{m_j,j}\). It is shown that \(\{\varphi_{m_1, \dots m_k}\}\) form a complete ONS in \(L^2(C_{a,b}[0,T], \mu)\) (Theorem 4). The second part of the paper is concerned with integral transforms of generalized Fourier-Hermitian functions \(F\) defined on the space of complex valued functions with real- and imaginary part belonging to \(C_{a,b}[0,T]\): For fixed \((\gamma, \beta)\) in \(\mathbb{C}^2\) define \(F_{\gamma,\beta} F(y) := \int F(\gamma x + \beta y) d\mu(x)\), generalizing the (classical) Fourier-Wiener transform of \textit{R. H. Cameron} and \textit{W. T. Martin} [Duke Math. J. 12, 489--507 (1945; Zbl 0060.27502); ibid. 14, 99--107 (1947; Zbl 0029.40002)].