2
$\begingroup$

Let $E$ be an elliptic curve over a number field $k$. We define the affine curve $C := E \backslash \{p_1,...,p_n\}$ by removing a finite number of points from $E$. Here, I would like to declare that I am not sure if its matters whether or not these points are $k$-points.

Then $C$ is a hyperbolic curve over $k$, i.e., it is the complement of $n$ points in a projective algebraic curve of genus $g$, such that $2g-2+n>0$. By the paper The Brauer-Manin obstruction for integral points on curves by Harari and Voloch, the paragraph preceding Lemma 2.1 states that affine curves whose complement in $\mathbb{P}^1_k$ is a reduced divisor of degree $\geq 2$ can be embedded in their generalized Jacobian, which in this case is an algebraic torus. Such affine curves include genus zero hyperbolic curves, i.e., open subsets of $\mathbb{P}^1_k$ with at least three points removed.

Question. Morphisms of varieties into their generalized Jacobians can be found in many literatures, but apart from the abovementioned example, I can't seem to find any where these morphisms are in fact embeddings. Does anyone know of some useful references, at least for our particular case $C$?

In general, the generalized Jacobian of $C$ is a semi-abelian variety $S$. One would suspect that $S$ could be the extension of the Jacobian variety $J_E$ of $E$ by some torus dependent on $n$, the degree of the divisor $D :=E\backslash C$. The closest thing I've managed to gather is that the "classical" Albanese variety of $E$ (which in this case is $J_E$) is the maximal abelian quotient of $S$. The kernel of the morphism $S \rightarrow J_E$ is some torus $T$ who dimension is the rank of the subgroup of divisors of $E \times _k \bar{k}$, supported on $D \times _k \bar{k}$, which are algebraically equivalent to zero.

EDIT. Here are some background on Albanese schemes, I'm referring to the paper Duality of Albanese and Picard 1-motives by Ramachandran. Given a connected scheme $X$, we have a universal morphism $u_X:X \rightarrow A_X$ from $X$ into its Albanese scheme $A_X$. The neutral component $A^0_X$ is called the generalized Albanese variety defined by Serre. If $X$ is proper, then $A^0_X$ is an abelian scheme.

In the case where $X$ is a normal integral curve, there exists a unique normal projective curve $X'$ containing $X$ as an open dense subscheme. Let $D := X'\backslash X$. The semiabelian variety $A^0_X$ is the generalized Jacobian of $X'$ (defined by Rosenlicht) corresponding to modulus $D$.

$\endgroup$
5
  • $\begingroup$ The Albanese (Abelian) variety is a birational invariant, hence extends to quasi-projective varieties. Since the Albanese morphism from $C$ to its Albanese variety $E$ is already an immersion, every morphism from $C$ to a semi-Abelian variety $S$ that factors the Albanese morphism is also an immersion. Are you asking whether the morphism from $C$ to $S$ is proper? $\endgroup$ Commented Jun 2, 2022 at 18:31
  • $\begingroup$ @JasonStarr If I'm not mistaken, the Albanese variety of $E$ has to coincide with $J_E$, and so you're implying that we have an immersion $C \rightarrow J_E$? I've included a portion on Albanese schemes, and realised I'm confused about one thing: from my reference and the book on algebraic curves by Serre, there was never any mention about the generalized Jacobian of an affine curve, the curve in question was always assumed to be proper. But yet the paper by Harari and Voloch clearly referred to that of an affine curve. Do you know where I possibly misunderstood? $\endgroup$
    – oleout
    Commented Jun 3, 2022 at 5:52
  • $\begingroup$ @JasonStarr Ignore the second question on generalized Jacobian, think I just figured that bit out. $\endgroup$
    – oleout
    Commented Jun 3, 2022 at 5:59
  • $\begingroup$ Yes, the identity map from $E$ to $E$ is an immersion, thus its restriction to the Zariski open subset $C$ is also an immersion, $C\to J_E$. $\endgroup$ Commented Jun 3, 2022 at 10:33
  • $\begingroup$ @JasonStarr I stupidly forgot that $E$ is its own Jacobian... Ok if $C := E\backslash \{p_1,p_2\}$, then I think it is not hard to show that we have an exact sequence $$1 \rightarrow T \rightarrow J_m \rightarrow E \rightarrow 1$$ where $T$ is a one-dimensional torus and $J_m$ is the generalized Jacobian of $E$ corresponding to the modulus $\{p_1,p_2\}$. Thus we have a universal morphism $f:E \rightarrow J_m$ such that $f|_C$ is regular. Do you know if this restriction is also a closed immersion? $\endgroup$
    – oleout
    Commented Jun 3, 2022 at 11:57

0

You must log in to answer this question.

Browse other questions tagged .