Extremal cases for the log canonical threshold (Q2255244): Difference between revisions

Let \(\Omega\) be a domain in \(\mathbb C^n\) such that \(0\in \Omega\), let \(\mathcal {PSH}(\Omega)\) be the class of plurisubharmonic functions defined on \(\Omega\), and let \(\mathcal E\) be the Cegrell class of plurisubharmonic functions on \(\Omega\) for which the complex Monge-Ampère operator \((dd^c \cdot)^n\) is well defined. For \(u\in \mathcal {PSH}(\Omega)\) let \(\lambda_u\) denote the integrability index of \(u\) at \(0\), \(\lambda_u=\inf\{\lambda>0; e^{-\frac u{\lambda}}\in L^2_{\text{loc}}(0)\}\), and let \(c_u=\lambda^{-1}_u\) denotes the log canonical threshold of \(u\). Let \(e_k(u)\) be the Lelong numbers of the current \((dd^cu)^n\) at \(0\) \[ e_k(u)=(dd^cu)^{k}\wedge (dd^c\log |z|)^{n-k}(0), \;\;1\leq k\leq n. \] \textit{J.-P. Demailly} [in: Complex analysis and digital geometry. Proceedings from the Kiselmanfest, Uppsala, Sweden, May 2006 on the occasion of Christer Kiselman's retirement. Uppsala: Univ. Uppsala. 131--143 (2009; Zbl 1209.32024)] and \textit{A. Zeriahi} [ibid. 144--146 (2009; Zbl 1206.32018)] proved that for \(u\in \mathcal E\) \[ c_u\geq ne_n(u)^{-\frac 1n}.\tag{*} \] The aim of this note is to give a description of plurisubharmonic functions with fixed Monge-Ampère mass and smallest possible log canonical threshold (i.e., with equality in (*)). The author proves that for any \(u\in \mathcal E\) near \(0\) it holds \(c_u=ne_n(u)^{-\frac 1n}\) if and only if its greenification \(g_u\) for some bounded hyperconvex domin \(\Omega\) satisfies \(g_u=e_1(u)\log |z|+O(1)\), as \(z\to 0\), where \[ g_u=\sup \big\{v\in \mathcal {PSH}(\Omega); v\leq u+O(1) \text{ near } 0\big\}^*. \] The proof follows from the following inequality due to \textit{J.-P. Demailly} and \textit{H. H. Phạm} [Acta Math. 212, No. 1, 1--9 (2014; Zbl 1298.14006)]: for \(u\in \mathcal E\), \(e_1(u)>0\) it holds \[ c_u\geq \sum_{k=1}^n \frac {e_{k-1}(u)}{e_k(u)}. \]