Proposition 10.16.3.5D137EECD5CC The quantities in (10.16.1) and (10.16.2) are finite, and the latter is independent of the Kähler resolution. If $\eta \in b$ is smooth, $\kappa $ is Kähler and $C>0$ satisfies $C\kappa -\eta >0$, then
A4CFA8A325BE
10.16.4
\begin{equation} \mathcal R_ Z(b+t\{ \kappa \} ) \leq (C+t)^ k\int _ Z\kappa ^ k \qquad (t\geq 0). \end{equation}
Let $\mu :X'\to X$ be a modification and $Y'$ the proper transform of a subvariety $Y$ that meets the locus where $\mu $ is an isomorphism. Then
\[ \mathcal R(\mu ^*a;Y')=\mathcal R(a;Y). \]
Proof.
Step 1: Kähler forms. Let $\lambda $ be a Kähler form on $Z$ and $q\in \mathrm{PSH}(Z,\lambda )$ with $q\leq 0$. The truncations $q_ j=\max (q,-j)$ have mass $\int _ Z\lambda ^ k$ by theorem 11.1.7, or by Stokes' theorem when $k=1$. Restricting their Monge–Ampère measures to $\{ q>-j\} $ and passing to the increasing limit gives
\[ \int _ Z\langle (\lambda +\mathrm{dd}^{\mathrm{c}}q)^ k\rangle \leq \int _ Z\lambda ^ k. \]
The smooth current $\lambda $ attains equality.
Step 2: the mass bound. Put $A=C\kappa -\eta $. For a positive current $S$ in $b+t\{ \kappa \} $, the current $S+A$ is positive and belongs to $(C+t)\{ \kappa \} $. Since local non-pluripolar products are positive and multilinear, $\langle S^ k\rangle \leq \langle (S+A)^ k\rangle $, and Step 1 applied to the Kähler form $(C+t)\kappa $ yields (10.16.4). In particular $\mathcal R_ Z(b)$ is finite. The definition involves only the class, so the choice of its smooth representative or of the comparison form does not change the supremum.
Step 3: modifications. Let $\rho :W\to Z$ be a modification between compact Kähler manifolds. By proposition 5.43.18, every positive current in $\rho ^*b$ is the pullback of a unique positive current in $b$. Their non-pluripolar measures correspond on the locus where $\rho $ is an isomorphism, and neither charges the exceptional analytic sets. Hence their total masses agree and
\[ \mathcal R_ W(\rho ^*b)=\mathcal R_ Z(b). \]
Applying this identity on a common Kähler refinement shows that $\mathcal R(a;Y)$ is independent of the resolution. The same argument compares resolutions of $Y$ and $Y'$, since their maps to $X$ agree on a dense open set and therefore everywhere on a common refinement.
$\square$
Comments
Corrections, questions, and mathematical discussion are welcome. Sign in with GitHub to comment. Comments are public.