0B216F0B911E
Fix a smooth positive probability volume form $\mathrm{d}V$ on $X$, used for $L^1$ norms. Consider a sequence of plurisubharmonic functions on a connected domain that is locally uniformly bounded above. By the local compactness theorem recalled in Theorem 5.10.4, it has a subsequence that either converges in $L^1_{\mathrm{loc}}$ or converges locally uniformly to $-\infty $. For normalized $\theta $-plurisubharmonic functions on $X$, a point at which the maximum is attained excludes the second alternative, and connectedness of $X$ propagates the first one across $X$.
The submean inequality also controls upper limits at moving points.
Lemma 5.25.1.45B3FA75B7BD Let $u_ j,u$ be quasi-plurisubharmonic functions on an open set $U$ in a complex manifold. Suppose that $u_ j\to u$ in $L^1_{\mathrm{loc}}(U)$ and that $\mathrm{dd}^{\mathrm{c}}u_ j\geq -\beta $ for a fixed smooth real $(1,1)$-form $\beta $ on $U$. If $x_ j\to x$ in $U$, then
\[ \limsup _{j\to \infty }u_ j(x_ j)\leq u(x). \]
Moreover, the tail envelopes $h_ j=(\sup _{k\geq j}u_ k)^*$ decrease to $u$.
Proof.
On a coordinate ball about $x$, add a common smooth correction $q$ so that $u_ j+q$ and $u+q$ are plurisubharmonic. For a fixed small radius, the submean inequality bounds $(u_ j+q)(x_ j)$ by the average of $u_ j+q$ over the ball of that radius centered at $x_ j$. As $j\to \infty $, these averages converge to the average of $u+q$ over the ball of the same radius centered at $x$: the contribution of $u_ j-u$ tends to zero in $L^1$, and the integral of the fixed locally integrable function $u+q$ over a ball depends continuously on its center. Letting the radius decrease to zero proves the inequality.
The same estimate gives local uniform upper bounds for the $u_ j$ with $j$ large. By coefficient freezing and the local upper-envelope theorem, the tail envelopes are therefore quasi-plurisubharmonic, with the same lower Hessian bound. They decrease with $j$. If their limit exceeded $u(x)$ at some point $x$, the definition of the regularization would give indices $k_ j\geq j$ and points $x_ j\to x$ contradicting the first assertion for the subsequence $(u_{k_ j})$. On the other hand, an almost everywhere convergent subsequence of $(u_ j)$ shows that $h_ j\geq u$ almost everywhere, hence everywhere by lemma 5.24.3. Thus $h_ j\downarrow u$.
$\square$
Theorem 5.25.2.FC59B7D406A3 For every smooth real $(1,1)$-form $\theta $ on $X$, the family
\[ \mathcal K_\theta =\{ u\in \mathrm{PSH}(X,\theta ):\sup _ Xu=0\} \]
is compact in $L^1(X,\mathrm{d}V)$. In particular, there is a constant $C_\theta $ such that
\[ \int _ X|u|\, \mathrm{d}V\leq C_\theta \quad (u\in \mathcal K_\theta ). \]
Proof.
Let $(u_ j)$ be a sequence in $\mathcal K_\theta $. By upper semicontinuity, there are points $x_ j$ with $u_ j(x_ j)=0$; passing to a subsequence, we may assume that $x_ j\to x$. Cover $X$ by finitely many connected coordinate balls, and on a slightly larger ball around each of them choose a fixed smooth function $r_ i$ such that every $u_ j+r_ i$ is plurisubharmonic; such corrections exist because $\theta $ is smooth. Since $u_ j\leq 0$, these local functions are uniformly bounded above.
On a ball containing $x$ in its interior, locally uniform convergence to $-\infty $ is impossible, because the values $(u_ j+r_ i)(x_ j)=r_ i(x_ j)$ stay bounded below. Local compactness therefore gives a subsequence converging in $L^1$ on this ball. If a second ball meets the first, local compactness on the second ball cannot yield convergence to $-\infty $ on compact subsets of the intersection, since the first limit is finite almost everywhere there. Thus a further subsequence converges in $L^1$ on the second ball. Since $X$ is connected, the intersection graph of the finite cover by connected balls is connected. Proceeding along this graph and passing to subsequences finitely many times, we obtain local $L^1$ convergence on every ball, and a partition of unity then gives convergence in $L^1(X)$ to a function $u$.
The inequalities $\theta +\mathrm{dd}^{\mathrm{c}}u_ j\geq 0$ pass to the limit in the sense of distributions. By proposition 5.22.2, $u$ has a $\theta $-plurisubharmonic representative, still denoted by $u$. Since $u_ j\leq 0$, we have $u\leq 0$ almost everywhere, hence everywhere by lemma 5.24.3.
By lemma 5.25.1, $0=\lim _ j u_ j(x_ j)\leq u(x)$. Thus $u(x)=0$, and the limit belongs to $\mathcal K_\theta $. This proves sequential compactness, which is compactness in the metric space $L^1(X)$. The norm bound follows, since a compact subset of $L^1(X)$ is bounded.
$\square$
Corollary 5.25.3.0AD144F69807 Let $(u_ j)$ be a decreasing sequence in $\mathrm{PSH}(X,\theta )$ with uniformly bounded $L^1$ norms. Then its pointwise limit $u$ belongs to $\mathrm{PSH}(X,\theta )$; in particular, $u$ is not identically $-\infty $. Moreover, $u_ j\to u$ in $L^1$.
Proof.
On each connected coordinate ball, after a common smooth correction, the limit of a decreasing sequence of plurisubharmonic functions is either plurisubharmonic or identically $-\infty $. Fatou's lemma, applied to the negative parts, excludes the second alternative because of the $L^1$ bound. Hence $u$ is quasi-plurisubharmonic, and the inequality $\theta +\mathrm{dd}^{\mathrm{c}}u\geq 0$ passes to the decreasing limit. Finally, the nonnegative functions $u_1-u_ j$ increase to $u_1-u$, so monotone convergence gives $u_ j\to u$ in $L^1$.
$\square$
Comments
Corrections, questions, and mathematical discussion are welcome. Sign in with GitHub to comment. Comments are public.