Plurisubharmonic approximation and boundary values of plurisubharmonic functions (Q2338695)

Let \(X\subset \mathbb C^n\) be a compact set. A Jensen measure with barycenter \(z\in X\) (\(\mathcal J_z(X)\)) is a regular Borel probability measure \(\mu\) with support in \(X\) such that \(u(z)\leq \int u\,d\mu\) for every function \(u\) which is plurisubharmonic and continuous in a neighborhood of \(X\). A function \(u:X\to [-\infty,+\infty)\) is called plurisubharmonic on \(X\) (\(u\in \mathcal {PSH}(X)\)) if \(u\) is upper semicontinuous (\(u\in \mathcal {USC}(X)\)) and \[ \forall \, z\in X\; \forall \, \mu\in \mathcal J_z(X)\quad u(z)\leq \int u\,d\mu. \] In this paper the following problem is studied. In a bounded domain \(\Omega\subset \mathbb C^n\), which upper semicontinuous functions on \(\overline \Omega\) and plurisubharmonic on \(\Omega\) are plurisubharmonic on \(\overline \Omega\)\,? To answer this question a new class of domains is introduced. A bounded domain \(\Omega\subset\mathbb C^n\) is called \textit{P-hyperconvex} if there exists \(u\in \mathcal {PSH}(\overline \Omega)\cap \mathcal C(\overline \Omega)\) such that \(u=0\) on \(\partial \Omega\). P-hyperconvex domains are hyperconvex but not necessarily strictly hyperconvex. It is proved also that every bounded pseudoconvex domain with \(\mathcal C^1\) boundary is P-hyperconvex. Moreover, P-hyperconvex domains can be also characterized in terms of Jensen measures. Namely, \(\Omega\) is P-hyperconvex if and only if \[ \forall \, z\in \partial \Omega \; \forall \, \mu\in \mathcal {J}_z(\overline \Omega) \quad \text{supp} \mu \subset \partial \Omega. \] The main theorem of this paper is the following. Let \(\Omega\) be a P-hyperconvex domain and let \(u\in \mathcal {PSH}(\Omega)\cap \mathcal {USC}(\overline \Omega)\), then \(u\in \mathcal {PSH}(\overline \Omega)\) if there exists \(v\in \mathcal {PSH}(\overline \Omega)\) such that \(u=v\) on \(\partial \Omega\). In other words, a plurisubharmonic function on \(\Omega\) is plurisubharmonic on \(\overline \Omega\) if it has ``right'' boundary values. This leads to the following Dirichlet problem: for a given \(f\in \mathcal {USC}(\partial \Omega)\), find \(u\in \mathcal {PSH}(\overline \Omega)\) such that \(u=f\) on \(\partial \Omega\). It is proved in the paper that the above problem for P-hyperconvex domains can be solved if and only if \(f\in \mathcal {PSH}(\partial \Omega)\).