On the distributions of smooth functions on infinite-dimensional spaces with measures (Q403911)

This paper concerns the absolute continuity of an image measure by a map \(f\). Let \(X\) be a locally convex space and \(\mu\) be a Radon probability measure on \(X\). The main result of the paper is as follows. Theorem 1. Let \(X\), \(\mu\) be as above, \(\mu\) be continuous along vectors for a countable set \(S\), and let \(f\) be a \(\mu\)-measurable function on \(X\) such that all partial derivatives \(\partial_{h_1}\cdots\partial_{h_n}f\) exist for all \(h_1,h_2,\dots,h_n\) in the linear span of \(S\). Let \(E= \{x:\exists h_1,h_2,\dots, h_n\in S,\,\partial_{h_1}\cdots\partial_{h_n} f\neq 0\}\). Then the measure \(\mu|_E\circ f^{-1}\) is absolutely continuous. It is possible to discuss the above result in more detail if \(\mu\) is Gaussian-Radon measure. One of these results is given in the following. Corollary 1. Let \(\mu\) be a centered Radon-Gaussian measure on \(X\), \(f\in W^{p,\infty}(\mu)\) and \(E= \bigcup^\infty_{n=1} \{x: D^n_H f(x)\neq 0\}\), where \(D^n_H\) is the Sobolev derivative along the Cameron-Martin space \(H\). Then the measure \(\mu|_E\circ f^{-1}\) is absolutely continuous. The same is true for \(f\) of class \(W^{p, \infty}_{\text{loc}}(\mu)\), where \(W^{p,\infty}(\mu)\) \((W^{p,\infty}_{\text{loc}}(\mu))\) is the Sobolev class (local Sobolev class), respectively.