Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

I start my question with a definition and some motivation.

Let $M$ be a symplectic manifold. A subbundle $P\subset TM^{\mathbf{C}}$ of the complexified tangent bundle is called a complex polarization if

  1. $P$ is Lagrangian, i.e. Maximal isotroic, dim$P_m=n$, $\forall m\in M$, and $\omega(P,P)=0$

  2. P involutive, i.e. if $X,Y\in P$ then $[X,Y]\in P$

  3. dim$P\cap\bar P \cap TM$ is constant

Now, Introduce an hermitian form on $P$ defined by

$$b(X,Y)=i\omega(X,\bar Y).$$

Note that if $P$ is real, then $b$ is vanishes identically on $P$. Consequently, $b$ projects onto a non-degenerate form on the quotient $P/{(P\cap \bar P)}$ and we denote it by $\bar b$. $P$ is said to be of type $(r,s)$ if and only if $\bar b$ has signature $(r,s)$ i.e. its matrix is $$diag(\underbrace{1,1...,1}_{r},\underbrace{-1,-1...,-1}_{s} )$$ for $0\le r+s=n-dim_{\mathbf C }P\cap \bar P$ . Then, $P$ is said to be positive if $s=0$. In the case if $r=s=0$ then $P$ is real

Let $(M,\omega, J)$ be a compact Kahler manifold with positive-definite polarization $P$, and $(L,\nabla)$ be prequantum data. Let $$M_{quantum} = \left\{s\in \Gamma(L) \vert \nabla_Xs=0 , \forall X\in \bar P \right\}$$ Then I can not see why $M_{quantum}$ is fnite-dimensional?. I am looking for a referrence for finding a proof for this assertion "the space of square integrable holomorphic sections is closed."

share|improve this question
Can you define precisely what you mean by Lagrangian and involutive? –  Liviu Nicolaescu Dec 31 '13 at 14:31
Involutive, means, if $X,Y\in P$ then $[X,Y]\in P$. and Lagrangian means maximal isotropic –  Hassan Jolany Dec 31 '13 at 14:37
Maximal isotropic with respect to what bilinear form? –  Liviu Nicolaescu Dec 31 '13 at 14:56
I am asking all these question because there is another notion of polarization used in a similar context. In that case $P\cap \bar{P}=0$, by definition. –  Liviu Nicolaescu Dec 31 '13 at 15:15
in the case $P\cap \bar{P}=0$ we say kahler polarization. and is special case of polarization –  Hassan Jolany Dec 31 '13 at 15:53

1 Answer 1

up vote 1 down vote accepted

$\newcommand{\bC}{\mathbb{C}}$ $\newcommand{\bR}{\mathbb{R}}$ $\DeclareMathOperator{\Hom}{Hom}$ Suppose that $V$ is a finite dimensional real space equipped with an almost complex structure $J$. Let $P\subset V^{\bC}=V\otimes_{\bR}\bC$ be a complex vector subpace.

Then $P^*=\Hom_{\bC}(P,\bC)$ is a quotient of $V_{\bC}^*=\Hom_{\bC}(V^{\bC},\bC)=\Hom_{\bR}(V,\bC)$ . We denote by $\Pi_P$ the projection $V_{\bC}^*\to P^*$. We define $\Pi_{\bar{P}}$ similarly. Now observe that $V^*_{\bR}=\Hom(V,\bR)$ is a real subspace of $V_{\bC}^*$. For $\xi\in V^*_{\bR}$ we set $\xi_{\bar{P}}:=\Pi_{\bar{P}}\xi.$

If the induced real linear map $\xi\mapsto \xi_{\bar{P}}$ is injective, then the answer to your question is positive.

To see this consider the differential operator $\newcommand{\pap}{\partial_{\bar{P}}}$ $\pap^\nabla$ defined as the composition

$$ C^\infty(L)\stackrel{\nabla}{\to} C^\infty(T^*M\otimes_{\bR} L)\stackrel{\Pi_{\bar{P}}}{\to} C^\infty( \bar{P}^*\otimes L). $$


$$M_{quantum}=\ker \pap^\nabla=\ker (\pap^\nabla)^*\pap^\nabla. $$

If the map $\xi\mapsto\xi_{\bar{P}}$ is injective, then the operator $(\pap^\nabla)^*\pap^\nabla$ is elliptic and it has finite dimensional kernel on the compact manifold $M$.

share|improve this answer
In your solution you have assumed $P\subset V^{\bC}=V\otimes_{\bR}\bC$ while we tae $P$ as subbundle of $TM\otimes \bC$ –  Hassan Jolany Dec 31 '13 at 16:09
At every point $x$ in $M$ the fiber $P_x$ is a subspace of $V^{\mathbb{C}}$ where $V=T_xM$. –  Liviu Nicolaescu Dec 31 '13 at 16:39
Thanks a lot, for your nice answer –  Hassan Jolany Dec 31 '13 at 16:42
I am still waiting for second part of answer. –  Hassan Jolany Dec 31 '13 at 17:18
Once you have the corresponding ellipticity, the statement about the closedness is immediate: all holomorphic sections are smooth in particular square integrable. They form a finite dimensional subspace of a Hilbert space, hence it is a closed subspace. I reiterate my comment that your question needs to be more carefully formulated because the meaning of isotropic is trickier than you think. –  Liviu Nicolaescu Dec 31 '13 at 17:40

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.