Sur la décomposition de la trajectoire d'un processus de Lévy spectralement positif en son infimum. (On the path decomposition at the infimum for a spectrally positive Lévy process) (Q1185277)

Let \(X\) be the coordinate process on the space \(\Omega\) of càdlàg functions with a life time \(\zeta\) and the natural filtration \(({\mathcal F}_ t\), \(t\geq 0)\), \(\overline {X}_ t=\sup\{X_ s\), \(0\leq s\leq t\}\), \(\underline {X}_ t=\inf\{X_ s\), \(0\leq s\leq t\}\), \(\underline{\underline {X}}_ t=\inf\{X_ s\), \(t\leq s<\zeta\}\), \(\rho=\sup\{t\geq 0\): \(\underline{\underline {X}}_ t=\underline {X}_ \infty\}\), \(\tau(-x)=\inf\{t\geq 0\): \(X_ t<-x\}\). It is considered a probability measure \(P\) on \({\mathcal F}_ \infty\) such that \(X\) is a spectrally positive Lévy process starting from 0 and drifting to \(+\infty\). Denoting \(\alpha\) the unique positive solution of \(\psi(\alpha)=0\), where \(\psi\) is the cumulant function of \(X\), define the probability measure \(P^*\) by means of the equality \[ dP^*\mid_{{\mathcal F}_ t}=\exp\{-\alpha X_ t\}dP\mid_{{\mathcal F}_ t}, \qquad t\geq 0. \] It is proved that the probability distribution of the preinfimum process \((X_ t\), \(0\leq t<\rho)\) with respect to \(P\) coincides with the probability distribution of the process \((X_ t\), \(0\leq t<\tau(-\gamma))\) with respect to \(P^*\), where \(\gamma\) is independent of \(X\) and exponentially distributed with the parameter \(\alpha\) and the probability distribution of \(X- 2\underline{\underline {X}}^ c-{\mathcal J}\) with respect to \(P\) equals to \(P^*\), where \(\underline{\underline {X}}^ c\) denotes the continuous part of \(\underline{\underline {X}}\) and \[ {\mathcal J}_ t=\sum_{0<s\leq t}(X_ s-X_{s-})\text{\textbf{1}}_{\{\underline{\underline {X}}_ s>X_{s-}\}}. \] The known inverse Pitman's theorem is extended.