Riesz transform and integration by parts formulas for random variables (Q544522): Difference between revisions

Integration by parts formulae on the Wiener space enable to prove representation formulae for the distribution laws of variables. In particular, one can write a formula for the density of a variable \(F\), involving the expectation of \(\partial_i Q_d(F-x)\), where \(Q_d\) is the Poisson kernel on \(\mathbb R^d\setminus\{0\}\). A central point of interest in this work is the estimation of \(\mathbb E(|\partial_i Q_d(F-x)|^d)\). As a consequence, regularity and estimates for the density of \(F\) are obtained, and the set where the density is strictly positive is studied.