I've taken a small course in Riemann surfaces, and there is one part that I still don't understand (and I've been unable to find a reference that explains this rigorously and in detail).

It is about the construction of a Riemann surface associated to an analytic germ.

By analytic germ we mean a couple $(z_0, (a_n)_{n \in \mathbb{N}})$ where $z_0$ and the $a_i$ are in the complex plane, and the series $\sum a_n (z-z_0)^n$ has strictly positive (but typically finite) radius of convergence.

Then (without getting into technical details) the Riemann surface associated to such a germ is the connected component of that germ in the space of all analytic germs, equipped with a certain topology. I get that it defines a Riemann surface.

I also understand "intuitively" what it does on the classic examples : for example, the Riemann surface associated to $(1,\log)$ is a sort of "spiral surface", since when we turn around zero, we add (if turning counter-clockwise) a $2 i \pi$ to the principal determination (so there are countably many sheaves).

In the case of the germ of the square root at $z_0=1$, it is a two-sheaves surface.

However I have no idea how one would go to determine this properly. In this example, I know that this surface is biholomorphic to $\mathbb{C}$ in the case of the log but I have no idea how to prove that.

Can anyone either give me a detailed reference or a sketch of the proof (but with all the key arguments) ?


What you are describing is more or less Weierstrass's construction of the Riemann surface of a multi-valued function. It is a very intuitive construction, but, as you point out, not convenient to do concrete calculations with. If you want to explicitly describe the Riemann surface obtained by analytic continuation of a given germ, there is not much choice other than to go back to the way in which the particular germ was obtained - the behavior of a power series under analytic continuation can be very complicated and can't readily be predicted simply by looking at the coefficients $\{a_n\}$.

For example, if you want to show that $\log z$ increases by $2\pi i$ with every counter-clockwise turn of $z$ around the origin, it is much better to go back to the definition of $\log z$ than to start from the power series expansion of its principal branch around $z=1$, although a brute-force calculation is certainly possible. What is $\log z$? By definition, $\log z = \int_1^z dt/t$. Of course, this is only well-defined up to the choice of path from $1$ to $z$, which results in an ambiguity of $2\pi i n$ in the value of the function. What we see, in fact, is that $\log z$ is well-defined as a function on the universal cover of the punctured plane, as soon as we specify its value at any given point. This universal cover $E$ can be intuitively thought of as a helicoid, as you point out. And, as you say, it is biholomorphic to $\mathbb C$. The function $\log$, well-defined on $E$, is the isomorphism you are looking for.

  • $\begingroup$ well ok, but then how do yo see that the covering surface is simply connected ? also how would you do that if instead of log, which as you point out may be defined using a multivalues integral, the germ is something like $\sqrt{(z-a)(z-b)}$ ? also is it clear is this last example that all points different from a and b may be base points ? (i.e. the covered surface is the twice punctured plane) $\endgroup$ – Glougloubarbaki Nov 29 '11 at 20:25

The problem is that this is AFAIK treated in older texts, those which are still influenced by the first book by Weyl ("The Concept of a Riemann Surface").

You coud try e.g.

  • George Springer: "Introduction to Riemann Surfaces", chapter 3: "Riemann Surface of an Analytic Function"
  • $\begingroup$ thanks ! I've begun reading it and it seems it covers this in detail. $\endgroup$ – Glougloubarbaki Nov 29 '11 at 11:18

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.