Properties of convergence in Dirichlet structures (Q1950184)

Consider a Dirichlet structure with a square field operator \(\Gamma[X,Y]\) and let \(\Gamma[X]=(\Gamma[X_i,X_j])\) for \(X=(X_1,\dots,X_p)\). The Dirichlet structure is said to satisfy the energy image density criterion if the law of a multivariate variable \(X\) is absolutely continuous with respect to a Lebesgue measure on \(\mathbb{R}^p\) (\(\det \Gamma[X]>0\)). An old conjecture is that every Dirichlet structure satisfies this criterion. Here, it is proved that if \(\det \Gamma[X]>0\) almost surely the characteristic function of \(X\) tends to 0 at infinity (the law of \(X\) is a Rajchman measure). To this end, it is proved that if a sequence \(X_n\) converges to \(X\) for the Dirichlet structure, then the law of \(X_n\) converges in total variation to the law of \(X\); this result is applied to the case where \(X_n\) is obtained from \(X\) by convolution. Moreover, the result concerning convergence in total variation is also proved in Sobolev spaces.