Let $M=\mathbb{C}P^1$ be the complex projective space, $U_0=\{[z_0,z_1]:z_0\ne 0\}$, $U_1=\{[z_0,z_1]:z_1\ne 0\}$ be the coordinate charts and define $E=\{([z_0,z_1],x):x\in\operatorname{span}_{\mathbb{C}}[z_0,z_1]\}$ the complex line bundle over $M$ with the natural projection $\pi([z_0,z_1],x)=[z_0,z_1]$ and the bi-holomorphic maps:

$$\phi_0:\pi^{-1}(U_0)\rightarrow U_0\times\mathbb{C},\ ([z_0,z_1],a(1,z_1/z_0))\mapsto ([z_0,z_1],a)$$ $$\phi_1:\pi^{-1}(U_1)\rightarrow U_1\times\mathbb{C},\ ([z_0,z_1],b(z_0/z_1,1))\mapsto ([z_0,z_1],b)$$

The transition functions are then $f_0^1=z_0/z_1, f_1^0=z_1/z_0$.
Let $\lambda(D)$ be the line bundle determined by a divisor $D$. Now to be simple consider the case $p=[1,0]$ and $q=[0,1]$, let $f_0=z_1/z_0$ and $f_1=z_0/z_1$, then $f_0$ is well defined on $U_0$ such that $p$ is the only zero point thus the divisor $(f_0)=p=D\cap U_0$ (where $D=p+q$). Similarly we have $(f_1)=q=D\cap U_1$ so the transition function of $\lambda(D)$ is $f_0^1=f_0/f_1=z_1^2/z_0^2$, but not $z_1/z_0$ to be expected. (Dual bundles have inverse transition functions).

What's wrong with the above argument, thanks.


The hyperplane section bundle $\mathscr O(1)$ on $\mathbb P^1$ is defined by the transition functions $(g_{ij} = \frac {z_j}{z_i})_{ij}$. The pair of local functions $f_0 := \frac {z_1}{z_0}$ on $U_0$ and $f_1 := \frac {z_0}{z_1}$ on $U_1$ from your question satisfies on the intersection $U_0 \cap U_1 $the compatibility rule

$$f_0 = g_{01}^2 * f_1.$$

Hence $f = (f_0, f_1)$ is a section in $\mathscr O(2)$ - not $\mathscr O(1)$. It vanishes at the two points $p=(1:0) $ and $q=(0:1).$

Note. The bundle from your question is the tautological line bundle $E = \mathscr O(-1)$.

  • $\begingroup$ So $\lambda(p+q)$ is not the dual of the tautological line bundle? $\endgroup$ – Xipan Xiao Dec 31 '14 at 19:06
  • $\begingroup$ $\lambda(p+q)$ is the square of the dual of the tautological line bundle. $\endgroup$ – Jo Wehler Dec 31 '14 at 22:09
  • $\begingroup$ Sections of the line bundle $\mathscr O(n), n \in \mathbb N,$ are exactly the holomorphic sections with $n$ zeros, counted with multiplicities. They correspond bijectively to the homogeneous polynomials of degree $n$. $\endgroup$ – Jo Wehler Dec 31 '14 at 23:14

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.