In André Weil's dissertation, he considers two meromorphic functions $x,y$ on a complex curve.  He assumes every pole of $y$ is a pole of $x$, and its multiplicity as a pole of $y$  is no greater than its multiplicity as a pole of $x$.  Then he says there is some natural number $k$ and some complex $a\neq 0$ such that  $$ay^k+xP(x,y)+Q(y)=0$$ where $P(x,y)$ and $Q(y)$ are polynomials in $x,y$ with degree $<k$

I see the proof for genus 0 curves. There the field of meromorphic functions is $\mathbb{C}(z)$ and one-variable polynomial algebra suffices (unless I've made a mistake). But I do not see it for other curves. Can someone tell me how it is done?

  • 1
    $\begingroup$ I assume that the curve is compact, then the field of meromorphic functions on it is finitely generated over $\mathbb{C}$ with transcendence degree $1$. Therefore any $x,y$ would satisfy a polynomial relation. It can be rewritten in the above form. (I'm guessing the symbol between $a$ and $0$ is $\not=$.) $\endgroup$ – Donu Arapura Dec 20 '19 at 3:05
  • $\begingroup$ @DonuArapura Yes he surely means compact, though writing in 1927 the closest he comes to saying so is to say he is using the "birational viewpoint." I expect you are right about the strategy he had in mind, and i will try to work through what he says using that idea. $\endgroup$ – Colin McLarty Dec 20 '19 at 4:34

Let us write the irreducible equation relating $x$ and $y$ as $$P_k(x)y^k+P_{k-1}(x)y^{k-1}+\ldots+P_0(x)=0.$$ Consider the Newton polygon (the graph of the smallest concave function $\phi$ with $\phi(j)\geq \deg P_j,\; 0\leq j\leq k$. Condition on the poles of $x$ and $y$ tells us that $P_k=\mathrm{const}$, and all slopes of this graph are $\geq -1.$ This implies that $$\deg P_j\leq m-j,\quad 0\leq j\leq k-1.$$ Now unite all constant terms of $P_j$ times powers of $y$ into the polynomial $Q(y)$, $\deg Q<k$, and the rest is $xP(x,y)$, where degree of $P$ with respect to $x$ and $y$ is at most $k-1$.

| cite | improve this answer | |
  • $\begingroup$ This is likely the right route. But I do not see how the condition on poles shows $P_k$ has no poles. The condition allows shared poles of $x$ and $y$ to have arbitrarily higher multiplicity in $x$ than $y$. I suspect the proof may need a step of multiplying through by some high enough power of $y$ to make all multiplicities of poles in $y^k$ surpass their multiplicity in $x$. $\endgroup$ – Colin McLarty Dec 20 '19 at 20:03
  • $\begingroup$ The proof is complete. If all poles of $y$ are poles of $x$ then $P_k$ is constant. Look at the properties of the Newton polygon. $\endgroup$ – Alexandre Eremenko Dec 20 '19 at 21:05
  • $\begingroup$ Can you say a little more about verifying that property of the Newton polygon? I am not familiar with calculating Newton polygons. $\endgroup$ – Colin McLarty Dec 20 '19 at 21:40
  • $\begingroup$ Algebraic equation wrt $y$ has $k$ solutions for generic $x$. If $P_k$ is not constant, it has a root at $x_0$ and as $x\to x_0$ some of these roots tend to $\infty$. Which means that at some point of the Riemann surface $x$ equals $x_0$ while $y$ has a pole. See any book on algebraic functions for the details. $\endgroup$ – Alexandre Eremenko Dec 20 '19 at 22:46
  • $\begingroup$ I see that argument. But the the conclusion on degrees in $x$ and $y$ seems to follow just by counting the multiplicity of any one pole. I do not see the role of the Newton polygon. $\endgroup$ – Colin McLarty Dec 21 '19 at 16:02

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.