The derivation of the Weierstrass elliptic function

1.7k Views Asked by At

I am wondering if any of you could point me to any books and/or lecture notes that explain the Weierstrass $\wp$ function for a self-studying student of elliptic curves and functions. I am interested in any resources that may give the history of the Weierstrass function and its derivation. I do understand the basics of lattices and doubly-periodic functions, but I am having trouble seeing the thought process that led to the creation of the function itself. I have tried searching through many books and while I have found several good ones, I haven't yet found one that shows the thought process that led to the Weierstrass $\wp$ function. (Sorry, but I am not sure how to format the letter that is traditionally associated with this function).

What resources are there to understand the derivation of $\wp$?

2

There are 2 best solutions below

7
On

HINT: here I give you a brief note, hoping that something you can help.

Knowing that all $L$-elliptic function (i.e. meromorphic and $L$-periodic) must have a finite order at least equal to $2$, the function $\wp$ gives a solution to the problem of construct a $L$-elliptic function taking exactly this minimum of twice all complex value $u$ in all fundamental parallelogram $P$ for the lattice $L$.

Assuming that such a function exists, it can have either just one pole of order 2 or two simple poles (two distinct kinds of these functions!); both exist in fact, the first one with a pole of order two are the $L$-functions $\wp$ of Weierstrass and the second one, the elliptic functions of Jacobi with two simple poles. The first function is defined by $$\wp(z)=\frac{1}{z^2}+\sum_{\omega\in L\setminus\{0\}}\left[\frac{1}{(z-\omega)^2}-\frac{1}{\omega^2}\right]$$ (a simpler form with $\sum \frac{1}{(z-\omega)^2}$ is useless because it is not convergent and the variation shown in $\wp$ has no other purpose than to ensure the necessary convergence).

If $P$ is the fundamental parallelogram of vertices $0,\omega_1,\omega_2,\omega_1+\omega_2$, the values $\wp(z_0)$ for $z_0=0,\frac{\omega_1}{2},\frac{\omega_2}{2},\frac{\omega_1+\omega_2}{2}$ are taken just once because $z_0$ is in each case a point of order two in $P$ (this because $0$ is a double pole with residue zero and the other three are double points since $\wp’=0$ and $\wp’’\ne 0$). All the other points $z\in P$ are simple and for all complex value $u\ne \wp(z_0)$ one has $\wp(z_1)=\wp(z_2)=u$ for certain unique $z_1$ and $z_2$ in $P$ which either are symmetric respect to $\frac {\omega_1+\omega_2}{2}$ (in whose case $z_1$ and $z_2$ are interior points of $P$) or are symmetric respect to $\frac{\omega_k}{2}$; $k=1,2$ (in whose case $z_1$ and $z_2$ are in the boundary of $P$, in the two sides assigned to $P$)

enter image description here

$\wp(z)$ has a derivative $\wp’(z)$ which is odd of order three and all $L$-elliptic function belongs to the field $\mathbb C(\wp, \wp’)$ of rational expressions of $\wp(z)$ and $\wp’(z)$ with complex coefficients; this field is a quadratic extension of the field $\mathbb C(\wp)$ formed by all the even $L$-elliptic functions.

$\wp$ and $\wp’$ are transcendental functions but are not algebraically independent and it is verify that $$(\wp’)^2=4\wp^3-g_2 \wp-g_3$$ where the called invariants of $\wp$ are defined by $$g_2=60\sum_{\omega\in L\setminus\{0\}}\frac{1}{\omega^4}$$ and $$g_3=140\sum_{\omega\in L\setminus\{0\}} \frac{1}{\omega^6}$$ Besides one has an “addition theorem” given by $$\wp(u+v)=\frac 14\left[\frac{\wp’(u)-\wp’(v)}{\wp(u)-\wp(v)}\right]^2-\wp(u)-\wp(v)$$ These two equalities allow us go from the transcendent to algebraic and this is the link between the elliptic functions and the cubic curves of genus 1.

0
On

I know it's too late but I was also stuck at finding references for the construction of the Weierstrass p function while reading [1] and here is what I ended up with.

  • [1] and Prof. Borcherds discuss for $\omega \in \Omega$ where $\Omega$ is a fundamental parallelogram representing a set of periods of a meromorphic function, $\sum_{\omega \in \Omega} \frac{1}{(z - \omega)^2}$ is not convergent (e.g., Theorem 3.9.2 of [1]) so they subtracted the residual part (See Prof.'s video), $\frac{1}{\omega^2}$, giving us $\sum_{\omega \in \Omega} \frac{1}{(z - \omega)^2} - \frac{1}{\omega^2}$. But, $\omega = 0$ for this series blow-up so we take this case out of the series by going back to the form of $\frac{1}{(z - \omega)^2}$ with $\omega = 0$, which leaves us $\frac{1}{z^2} + \sum_{\omega \in \Omega - {0}} \frac{1}{(z - \omega)^2} - \frac{1}{\omega^2}$.
  • Lecture note from Harvard Math 213 a discusses the use of Taylor polynomial in the section "Partial Fraction Expansion of Weierstrass p Function" though I couldn't quite follow it...

Hope this helps you or someone!

Reference

[1] Jones, Gareth A., and David Singerman. Complex functions: an algebraic and geometric viewpoint. Cambridge university press, 1987