Let $X$ be a smooth projective curve, so that we may define the group of divisors $\mathrm{Div}\ X$. For any $D\in\mathrm{Div}$, we let $L(D)$ be the vector space of functions $f\in k(X)$ such that $D+(f)$ is effective; we also let $0\in L(D)$. Next, define $l(D):=\dim_k L(D)$. We define $K$ as the canonical class of the curve $X$ (so $L(K)$ is defined up to isomorphism, and $l(K)$ is well defined).

I know that the concept of a morphism associated to a divisor exists, and I have heard about it, but I can't find it in Shafarevich, or in Gathmann, for example. As a result, what follows will be very vague.

I'm reasonably sure that there exists some rational map $\varphi_D:X\to\mathbb{P}^{l(D)-1}$, or possibly $\varphi_D:X\to\mathbb{P}^{l(K)-1}$ associated to $D$. This map has the property that if $D$ is very ample, that is to say, $l(D-p-q)=l(D)-2$ for all points $p,q\in X$, then $\varphi_D$ is an embedding (this is one application of Riemann-Roch).

What exactly is this map? How is it defined?

  • 1
    $\begingroup$ See the discussion after Theorem 3.3 in Shafarevich. It is surely in Gathmann as well. $\endgroup$
    – Hoot
    Aug 10, 2016 at 3:28

1 Answer 1


Given a divisor $D$, consider a basis $f_0, \dots, f_n$ of $L(D)$. Assuming the $f_i$ do not vanish at a common point $p \in X$, this gives a map $X \to \mathbb P^n$ (with $n+1=l(D)$) given by $p \mapsto [f_0(p): \dots, f_n(p)]$. Note choosing a different basis will correspond to composing the above map by an element of $PGL_n(k)$.

Rick Miranda has a section in his book talking about this sort of thing. It's about divisors and maps to projective space I think. His book is in the analytic setting, but it parallels the algebraic setting extremely closely.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .