Pome

13.18 Volume deficits and relative full massE8D1B2CE6F8F

E8D1B2CE6F8F

The upper-volume deficit measures how far one potential can be stretched away from another, and the positive lower volume of the less singular potential prevents the stretching envelopes from collapsing. We use the stretching envelopes of section 16.28. For $u,v\in \mathrm{PSH}(X,\theta )$ and $b\geq 1$, the competitor family $\mathcal A_ b(u,v)$ consists of all $h\in \mathrm{PSH}(X,\theta )$ with $b^{-1}h+(1-b^{-1})v\leq u$. When it is nonempty, its upper envelope is the $\theta $-plurisubharmonic function $Q_ b(u,v)=P_\theta (bu-(b-1)v)$, which again belongs to $\mathcal A_ b(u,v)$. For $\varphi \in \mathrm{PSH}(X,\theta )$, the relative envelope class $\mathcal E_{\mathrm{env}}(X,\theta ,\varphi )$ of definition 16.28.61 consists of all $u\in \mathrm{PSH}(X,\theta )$ with $u\preceq \varphi $ and $\mathcal A_ b(u,\varphi )\neq \varnothing $ for every $b\geq 1$.

Lemma 13.18.1.92ABF533A916 Let $u,v\in \mathrm{PSH}(X,\theta )$ satisfy $u\preceq v$, and put

\[ \delta =\underline{\operatorname {vol}}_\theta (v)>0, \qquad D=\overline{\operatorname {vol}}_\theta (v) -\overline{\operatorname {vol}}_\theta (u). \]

If $b>1$ and $b^ nD<\delta $, then $\mathcal A_ b(u,v)\neq \varnothing $; that is, the stretching envelope $Q_ b(u,v)$ is defined.

Proof. Adding constants to $u$ and $v$ only translates the competitors, so we may assume $v\leq 0$. Write $V=\overline{\operatorname {vol}}_\theta (v)$. Since $D<b^{-n}\delta $, the definition of the upper volume gives $w\in \mathrm{PSH}(X,\theta )$ with $w\sim u$ and

8F79AE2F6948 13.18.2
\begin{equation} V-\int _ X\mathrm{MA}_\theta (w)<b^{-n}\delta . \end{equation}

After a translation of $w$ we may also assume $w\leq v$. For positive integers $j$, put

\[ w_ j=\max (w,v-j),\qquad h_ j=Q_ b(w_ j,v). \]

The function $v-bj$ is a competitor for $h_ j$, and the defining convex inequality together with $w_ j\leq v$ gives $h_ j\leq v$. Thus $v-bj\leq h_ j\leq v$, so $w_ j\sim v$ and $h_ j\sim v$, and $h_ j$ decreases as $j$ increases. We claim that the decreasing limit $h$ of the $h_ j$ is not identically $-\infty $. If it were, theorem 5.25.2 would give $\sup _ Xh_ j\to -\infty $.

By theorem 15.8.6, the measure of $h_ j$ is carried by the contact set $C_ j=\{ b^{-1}h_ j+(1-b^{-1})v=w_ j\} $. Positivity of the mixed products and ordered contact for $b^{-1}h_ j+(1-b^{-1})v\leq w_ j$ give

\[ \mathrm{MA}_\theta (h_ j) \leq b^ n\mathbf1_{C_ j}\mathrm{MA}_\theta (w_ j). \]

Fix a positive integer $k$. If $h\equiv -\infty $, then $h_ j\leq -bk$ everywhere for all large $j>k$. Since $v\leq 0$, every point of $C_ j$ then satisfies $w_ j\leq -k$, hence $w\leq -k$. Moreover $w_ j=w$ on $\{ w>-k\} \subset \{ w>v-j\} $, so locality gives

\[ \begin{split} \delta \leq \int _ X\mathrm{MA}_\theta (h_ j) & \leq b^ n\int _{\{ w\leq -k\} }\mathrm{MA}_\theta (w_ j)\\ & \leq b^ n\left(V-\int _{\{ w>-k\} }\mathrm{MA}_\theta (w)\right). \end{split} \]

Here the first inequality uses $h_ j\sim v$, and the last uses $w_ j\sim v$. Letting $k\to \infty $ contradicts (13.18.2), since the finite non-pluripolar measure $\mathrm{MA}_\theta (w)$ gives no mass to the pole set of $w$. This proves the claim. Consequently $h\in \mathrm{PSH}(X,\theta )$, and $b^{-1}h+(1-b^{-1})v\leq w$. If $w\leq u+C$, then $h-bC\in \mathcal A_ b(u,v)$, which proves the lemma. $\square$

Proposition 13.18.3.169DBA6BAF3B Let $\pi :Z\to X$ be a smooth compact modification. There is a finite constant $C_\pi $, depending only on $\pi $ and $\theta $, such that, for $u,v,\delta ,D$ as in lemma 13.18.1,

\[ 0\leq \nu (u\circ \pi ,z)-\nu (v\circ \pi ,z) \leq C_\pi \min \left\{ 1,\left(\frac D\delta \right)^{1/n}\right\} \qquad (z\in Z). \]

The same estimate holds for generic Lelong numbers along every prime divisor on $Z$.

Proof. Let $C_\pi $ be the uniform bound for Lelong numbers of $\pi ^*\theta $-plurisubharmonic functions on $Z$ given by lemma 5.38.8. Since $u\preceq v$, the difference of Lelong numbers is nonnegative, and it is at most $\nu (u\circ \pi ,z)\leq C_\pi $. This proves the assertion when $D\geq \delta $.

If $D<\delta $, then by lemma 13.18.1 every $b>1$ with $b^ nD<\delta $ admits some $h\in \mathcal A_ b(u,v)$. Pulling back the defining convex inequality and taking Lelong numbers gives

\[ \nu (u\circ \pi ,z)-\nu (v\circ \pi ,z) \leq b^{-1}\bigl (\nu (h\circ \pi ,z)-\nu (v\circ \pi ,z)\bigr ) \leq b^{-1}C_\pi . \]

For $D>0$, let $b\uparrow (\delta /D)^{1/n}$; for $D=0$, let $b\to \infty $. Taking generic values along a prime divisor proves the divisorial assertion. $\square$

At zero deficit every finite stretch is possible. This leads to a second description of the relative envelope class. Fix a reference potential $\varphi \in \mathrm{PSH}(X,\theta )$.

Definition 13.18.4.FC98C471363B The relative full-mass class of $\varphi $ is

\[ \mathcal E_{\mathrm{vol}}(X,\theta ,\varphi ) =\{ u\in \mathrm{PSH}(X,\theta ):u\preceq \varphi , \ \overline{\operatorname {vol}}_\theta (u) =\overline{\operatorname {vol}}_\theta (\varphi )\} . \]

Conversely, convex combinations turn the envelope condition into equality of volumes.

Theorem 13.18.5.D73DFBAE1EC1 Suppose $\underline{\operatorname {vol}}_\theta (\varphi )>0$. Then

\[ \mathcal E_{\mathrm{env}}(X,\theta ,\varphi ) =\mathcal E_{\mathrm{vol}}(X,\theta ,\varphi ). \]

Every member $u$ of this class also satisfies $\underline{\operatorname {vol}}_\theta (u) =\underline{\operatorname {vol}}_\theta (\varphi )$.

Proof. First let $u\in \mathcal E_{\mathrm{env}}(X,\theta ,\varphi )$. For each $a\in (0,1)$, a member of $\mathcal A_ b(u,\varphi )$ with $b=(1-a)^{-1}$ gives $h\in \mathrm{PSH}(X,\theta )$ such that $a\varphi +(1-a)h\leq u$. By theorem 13.17.1, proposition 13.17.2 and homogeneity, for either volume,

\[ a^ n\operatorname {vol}_\theta (\varphi ) \leq \operatorname {vol}_\theta (a\varphi +(1-a)h) \leq \operatorname {vol}_\theta (u) \leq \operatorname {vol}_\theta (\varphi ). \]

Letting $a\to 1$ proves that both volumes of $u$ equal those of $\varphi $; this direction does not use positive lower volume.

Conversely, let $u\in \mathcal E_{\mathrm{vol}}(X,\theta ,\varphi )$. Then $D=0$ in lemma 13.18.1 with $v=\varphi $, so $\mathcal A_ b(u,\varphi )\neq \varnothing $ for every $b>1$. For $b=1$ the family contains $u$. This is the defining condition of $\mathcal E_{\mathrm{env}}(X,\theta ,\varphi )$. $\square$

Cite this page

Comments

Corrections, questions, and mathematical discussion are welcome. Sign in with GitHub to comment. Comments are public.