Let $X$ be a compact, genus $g$ Riemann surface (given as the desingularization and compactification of a plane algebraic curve), $J(X)$ its Jacobian, and $A : X \to J(X)$ the Abel map

$$A(P) = \left( \int_{P_0}^P \omega_1, \ldots, \int_{P_0}^P \omega_g \right),$$

where $P_0 \in X$ is some fixed place and $\{\omega_1, \ldots, \omega_g\}$ are the normalized basis of holomorphic differentials on $X$.

My goal is to solve the following equation for $K$, the Riemann constant vector:

Theorem: For any canonical divisor $\mathcal{C} \in \text{Div}(X)$, $$2K \equiv - A(\mathcal{C})$$

In general, there are $2^{2g}$ possible solutions to this equation: one for each half period $h_j = \alpha_j + \Omega \beta_j$ where $[I \; \Omega]$ is the period matrix of $X$ (i.e. $J(X) = \mathbb{C}^g / \Lambda$ where $\Lambda = \mathbb{Z}^g + \Omega \mathbb{Z}^g$) and the vectors $\alpha_j,\beta_j \in \{0, 1/2\}^g$. Given a canonical divisor, my current approach is to construct $2^{2g}$ vectors

$$K_j := h_j - \tfrac{1}{2}A(\mathcal{C})$$

and test them against the following theorem:

Theorem: Let $\theta : J(X) \times \mathfrak{h}_g \to \mathbb{C}$ be the Riemann theta function. Then $$\theta(W, \Omega) = 0$$ if and only if $$W = A(\mathcal{D}) + K$$ for some degree $g-1$ effective divisor $\mathcal{D}$.

(That is, evaluate $\theta(A(\mathcal{D}) + K_j)$ for all $j$.) However, in my computational experiments I've found that the choice of correct $h_j$ appears to be unique! That is, only one of the $2^{2g}$ half periods leads to the correct Riemann constant vector.

Question: What sort of techniques / theorems are out there that would help in proving this uniqueness statement?

Follow-Up: Are there specific techniques that would lend themselves toward an algorithm for computing $K$ that does not scale exponentially with the genus of the Riemann surface? (Such as in this approach?)

I've done some reading on holomorphic line bundles and spin bundles and I think I may have found some necessary, but not sufficient conditions for the uniqueness of a solution to this problem. For example, I conjecture that the half period $h_j$ must be odd.


Your Answer

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

Browse other questions tagged or ask your own question.