I'm getting stuck with a proposition. Please can someone help me? Let $E$ be a holomorphic vector bundle with hermitian metric $h$ over a connected compact Kähler manifold $X$ with Kähler form $\omega$ and associated Kähler metric $g$. Let $s\in H^{0}(X,E)$ a non-zero section. Consider the current $T_{s}=\frac{i\partial\bar{\partial}log(h(s,s))}{2\pi}$. Let's write $F(h)$ for the curvature of the Chern connection respect to the metric $h$, and another current $R_s=\frac{ih(F(h)s,s)}{2\pi h(s,s)}$. How can I prove that $0 \leq T_{s}+R_{s}$ in the following sense: (regarding those currents as forms) the hermitian form associated to $T_{s}+R_{s}$ is semipositive definite. The Hermitian form associated to a (1,1)-real form $\delta$ is the one given in local coordinates by the matrix of coefficients of $-i\delta$ and is (semi-)positive if this matrix is.

  • $\begingroup$ First: about $F_s$, I think there's no need to speak of currents: as you say, it's just a (1,1)-form. Second: about $R_s$, as it is written it is just a smooth function on $M$, rather than a form (or a current). So, I don't understand how $T_s+R_s$ should be interpreted: as it's written even as a current it doesn't have homogeneus (bi)degree (or (bi)dimension). $\endgroup$ – Qfwfq May 14 '10 at 16:06
  • $\begingroup$ *sorry, I meant $T_s$ above. $\endgroup$ – Qfwfq May 14 '10 at 16:08
  • $\begingroup$ Also, it's not completely obvious to me that $R_s$ -as it's written- doesn't have singularities along the zero set of the section $s$ (but it doesn't sound absurd, since $F(h)$ is $C^{\infty}$-linear). $\endgroup$ – Qfwfq May 14 '10 at 16:11
  • $\begingroup$ Hi! Both $T_{s}$ and $R_{s}$ can have singularities because there's no hypothesis on $s$, this is why i said currents,anyway, you can assume that $s$ doesn't vanish at any point. $T_{s}$ and $R_{s}$ belong to $H^{1,1}(X)\cap H^2(X,\mathbb{R})$, $T_{s}$ because the operator $\partial\bar{\partial}$ sends smooth complex functions to (1,1) forms, $R_{s}$ because $F(h)$ is a $End(E)$-valued (1,1) form so $F(h)s$ is a $E$-valued (1,1) form and finally $h(F(h)s,s)$ is a (1,1) form. $\endgroup$ – Italo May 14 '10 at 17:43
  • 1
    $\begingroup$ Hi Henri! Take the hermitian product in the E-section part and the form part beheaves like scalar function. More explicitly in coordinates: choose a local trivialization so you have $UxV$ with $V$ complex vector and $U$ open set of ${\mathbb{C}}^{r}$, choose a frame ${v_i}_i$ on $V$, you can express the metric respect to this frame $(h_{lm})_{lm}$ so you have $h(dz^{\alpha}\otimes v_{j}, v_{k})=dz^{\alpha}h_{jk}$. Is it more clear? $\endgroup$ – Italo May 15 '10 at 8:42

This is a consequence of the generalized Lelong-Poincaré formula for vector bundles: denoting $D'$ the $(1,0)$ part of the Chern connection of $(E,h)$ and $\Theta_h(E)$ its Chern curvature, one has:

$$dd^c \log |s|^2=\frac{1}{|s|^2}\cdotp \left( |D's|^2-\frac{|\langle D's,s\rangle|^2}{|s|^2}- \langle \Theta_h(E)s, s\rangle\right)$$

and essentially by Cauchy-Schwartz inequality, one has $$|s|^2\cdotp |D's|^2-|\langle D's,s\rangle|^2 \ge 0.$$

  • 6
    $\begingroup$ See Lemma 2.10 of the lectur note of Takuro Mochizuki, KOBAYASHI-HITCHIN CORRESPONDENCE FOR TAME HARMONIC BUNDLES AND AN APPLICATION math.kyoto-u.ac.jp/preprint/2005/15mochizuki.pdf and take $θ=0$ $\endgroup$ – user21574 Jun 7 '17 at 20:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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