Donsker's delta function of Lévy process (Q5937968)

It is well known, that Donsker's delta function \(\delta(B(t- a))\), \(t> 0\), \(a\in \mathbb{R}\), defined by \[ \delta(B(t)- a)= {1\over 2\pi} \int^\infty_{-\infty} e^{ir(B(t)- a)}dr,\tag{1} \] where \(\delta\) is the Dirac delta function, and \(\{B(t), t>0\}\) is a Brownian motion in \(\mathbb{R}\), may be realized as a generalized functional in (Gaussian) white noise analysis. The Lévy white noise calculus scheme is adapted to define and study Donsker's delta function of the Lévy process. It is shown that Donsker's delta function \(\delta(X(t)- a)\) of a Lévy process \(X\) may be realized as generalized Lévy functional. It is assumed that \(\{X(t)\}\) is right continuous with left limit. After introduction of the spaces of test and generalized Lévy functionals in Section 4, the authors show that when \(\{B(t)\}\) in (1) is replaced by \(X(t)\) the identity still has sense as a Bochner integral, and as a consequence they define \(F(X(t)- a)\) for any tempered distribution \(F\). As an application of the result they prove a generalized Itô formula for a simple Lévy process.