A comparison principle for the log canonical threshold (Q367184): Difference between revisions

Let \(u\) be a plurisubharmonic function on a domain \(\Omega\subset\mathbb C^n\) containing \(0\). The log canonical threshold, or complex singularity exponent, \(c_u(0)\) of \(u\) at \(0\) is defined by \textit{J.-P. Demailly} and \textit{J. Kollár} [Ann. Sci. Éc. Norm. Supér. (4) 34, No. 4, 525--556 (2001; Zbl 0994.32021)] as the supremum of the numbers \(c\geq0\) for which the function \(e^{-2cu}\) is integrable in some neighborhood of \(0\) which may depend on \(c\). Assume that \(\Omega_j\), \(j\geq1\), are smoothly bounded domains such that \(\Omega_{j+1}\Subset\Omega_j\Subset\Omega\) for all \(j\) and \(\bigcap_{j\geq1}\Omega_j=\{0\}\). The author proves that if \(u,\,v\) are plurisubharmonic functions on \(\Omega\) such that \(u\geq v\) on \(\partial\Omega_j\) for all \(j\geq 1\) then \(c_u(0)\geq c_v(0)\). The proof is based on a lemma which states that \(c_u(0)=\lim_{j\to\infty}c_{\max(u,j\log\|z\|)}(0)\), hence the log canonical threshold of \(u\) can be computed using plurisubharmonic functions which are locally bounded away from \(0\).