trace map of jacobians of Riemann surface

29 Views Asked by At

I'am reading 'A first course in modular form'. On page 221 Given two compact riemann surface and holomophic map $h$ : $X \to Y$.

Suppose $h$ is a surjection of finite degree $d$, that $h$ is locally $e_x$-to-1 at each $x \in X$ as already mentioned here, and that the set of exceptional points in $X$ is defined as $\mathcal{E}=\{x \in X$ : $\left.e_x>1\right\}$, the finite set of points where $h$ is ramified. Let $Y^{\prime}=Y-h(\mathcal{E})$ and $X^{\prime}=h^{-1}\left(Y^{\prime}\right)$ be the Riemann surfaces obtained by removing the images of the exceptional points from $Y$ and their preimages from $X$. The restriction of $h$ away from ramification and its image, $$ h: X^{\prime} \longrightarrow Y^{\prime}, $$ is a $d$-fold covering map where $d=\operatorname{deg}(h)$. This means that every point $y \in Y^{\prime}$ has a neighborhood $\widetilde{U}$ whose inverse image is a disjoint union of neighborhoods $U_1 \ldots . . U_d$ in $X^{\prime}$ such that each restriction $h_i: U_i \longrightarrow \widetilde{U}$ of $h$ is invertible.

The trace map induced by $h$ is a linear map transferring differentials from $X$ to $Y$, $$ \operatorname{tr}_h: \Omega_{\mathrm{hol}}^1(X) \longrightarrow \Omega_{\mathrm{hol}}^1(Y) . $$

If $\delta$ is a path in $Y^{\prime}$ lifting to a path in $X^{\prime}$ and $h_i^{-1}$ is a local inverse of $h$ about $\delta(0)$ taking $\delta(0)$ to the initial point of the lift then $h_i^{-1}$ has an analytic continuation along $\delta$, the chaining together of overlapping local inverses culminating in the local inverse about $\delta(1)$ taking $\delta(1)$ to the terminal point of the lift. To define the trace, let $\omega \in \Omega_{\mathrm{hol}}^1(X)$. Suppose $y$ is a point in $Y^{\prime}$, so that $h$ has local inverses $h_i^{-1}: \widetilde{U} \longrightarrow U_i, i=1, \ldots, d$. The trace is defined on $\widetilde{U}$ as the sum of local pullbacks, $$ \left.\left(\operatorname{tr}_h \omega\right)\right|_{\tilde{U}}=\sum_{i=1}^d\left(h_i^{-1}\right)^*\left(\left.\omega\right|_{U_i}\right) . $$

This local definition pieces together to a well defined global trace on $Y^{\prime}$ because analytically continuing the local inverses $h_i^{-1}$ along any loop in $Y^{\prime}$ back to $y$ permutes them, leaving the trace unaltered. The trace extends holomorphically from $Y^{\prime}$ to all of $Y$ (Exercise 6.2.4). The trace dualizes to a linear map of dual spaces, $$ \operatorname{tr}_h^{\wedge}: \Omega_{\text {hol }}^1(Y)^{\wedge} \longrightarrow \Omega_{\text {hol }}^1(X)^{\wedge}, \quad \operatorname{tr}_h^{\wedge} \psi=\psi \circ \operatorname{tr}_h . $$

The reverse change of variable formula for any path $\delta$ in $Y^{\prime}$, $$ \int_\delta\left(h^{-1}\right)^* \omega=\int_{h^{-1} \circ \delta} \omega, \quad \omega \in \Omega_{\mathrm{hol}}^1(X), $$ is meaningful so long as $h^{-1}$ is understood to be some local inverse of $h$ at $\delta(0)$ analytically continued along $\delta$, making $h^{-1} \circ \delta$ the lift of $\delta$ starting at $h^{-1}(\delta(0))$. This reduces to a finite sum of the same result on coordinate patches, where it is just forward change of variable (6.3) with $h^{-1}$ in place of $h$. Summing the reverse change of variable formula over local inverses gives for paths $\delta$ in $Y^{\prime}$ $$ \int_\delta \operatorname{tr}_h \omega=\sum_{\text {lifts } \gamma} \int_\gamma \omega, \quad \omega \in \Omega_{\text {hol }}^1(X) . $$

Definition 6.2.3. The reverse map of Jacobians is the holomorphic homomorphism induced by composition with the trace, $$ h^J: \operatorname{Jac}(Y) \longrightarrow \operatorname{Jac}(X), \quad h^J[\psi]=\left[\psi \circ \operatorname{tr}_h\right] . $$

Writing elements of $\operatorname{Jac}(Y)$ as sums of integrations per Abel's Theorem, the summed reverse change of variable formula (6.6) shows that the reverse maps of Jacobians transfers integration modulo homology from $Y$ to $X$ by pulling back the limits of integration with suitable multiplicity, $$ h^J\left(\sum_y n_y \int_{y_0}^y\right)=\sum_y n_y \sum_{x \in h^{-1}(y)} e_x \int_{x_0}^x . $$

My question is that why $h^J$ looks as this. From the last second formula, it should be integration along all the lifts of a given curve from $y_0$ to $y$, and I don't see why the right part is the integration along a fix point $x_0$ to other $x$. I think $x_0$ shoule change as the lifts.