Throughout I would like to work over an algebraically closed field of characteristic 0 (so no separability issues), say $k$. My question is the following:
Do there exist two curves $X$ and $Y$ and a necessarily finite morphism $f:X \rightarrow Y$ such that $$[k(X):k(Y)] > \text{max}_{P \in X}\{e_P\}$$ where $k(X)$ denotes the function field of $X$ and $e_P$ denotes the ramification index of the map at a point $P \in X$?
My thoughts/attempts so far: Hurwitz's Theorem tells us that for a finite separable morphism of curves $f:X \rightarrow Y$ we have $$2g(X)-2=(\text{deg }f)(2g(Y)-2)+\text{deg }R$$ where $\text{deg }f=[k(X):k(Y)]$, $g(X)$ is the genus of $X$ and $R$ is the ramification divisor. That is, $\text{deg }R=\sum_{P\in X} (e_P-1)$.
Using this I can note that a simple example may come from a morphism between two genus 0 curves that gives a field extension of degree 3 and ramifies at 4 points. In this case each ramification index would have to be 2 and hence the condition would be satisfied. I thought I had an example of higher genus using the double cover of the projective line over $k$ by an elliptic curve, however at one of the four points of ramification the index would have to be at least equal to the field extension degree.
A related extension to this question is whether it is possible to have such a map that any possible ramification type occurs - that is, given two fixed genus values (necessarily $g(X)\geq g(Y)$ by Hurwitz) and $[k(X):k(Y)]$ fixed, can a morphism be constructed that ramifies at any allowable number of points with indices? Now I do not insist on the ramification indices all being smaller than the degree of the field extension. By allowable here I mean that deg $R$ is controlled by the fixed values by Hurwitz's Theorem, and so the greatest number of ramification points is deg $R$ each with index 2, but other partitions of deg $R$ with integers $\geq 2$ are possible.
Thanks,
Andrew
Take an étale covering $f:X\to Y$ of degree $n\gt 1$ (i.e. with $n$ sheets) between two elliptic curves ( such coverings exist for any $n $ ) .
You have $[k(X):k(Y)]=n$ but all the $e_P=1 $ and thus $$[k(X):k(Y)]=n\gt \text{max}_{P \in X}\{e_P\}=1$$
Edit: an explicit example.
To show how simple it is to construct such étale coverings, consider an arbitrary lattice $\Lambda =\mathbb Z\omega_1\oplus \mathbb Z\omega_2\subset \mathbb C$ and the related lattice $\Lambda' =\mathbb Z\omega_1\oplus \mathbb Z n\omega_2\subset \mathbb C$.
The morphism $\mathbb C/\Lambda \to \mathbb C/\Lambda':[z] \mapsto [nz]' $ is then an étale covering of degree $n$ .
(In the jargon of abelian varieties: an isogeny with kernel $\lbrace [0],[\frac {1}{n}\omega_1],[\frac {2}{n}\omega_1],...,[\frac {n-1}{n}\omega_1]\rbrace $)