#header,next=./PRELIM.html

June 2020

My sincere sympathies to all those suffering loss and hardship due to the COVID-19 pandemic.


Welcome to my web site.

This series of web pages are based on notes I made about 30 years ago when studying the theory of elliptic functions. In them we compute the $g_2$, $g_3$ invariants for several genus 1 curves of low degree. The $g_2$ and $g_3$ invariants are not quite intrinsic to the curve itself, but are uniquely defined once a specific holomorphic differential on the curve is chosen.

One of the original motivations for carrying out these computations was to answer what seemed like a simple question.


What is the differential equation for the general elliptic function of order 3 and how do you calculate the $g_2$ and $g_3$ invariants of the corresponding genus 1 curve?

The differential equation for elliptic functions $f$ of order 2 is

f'^2 = af^4 + bf^3 + cf^2 + df + e

and it is, more or less, the starting point of the theory of elliptic integrals. The associated curve of genus 1 is

y^2 = ax^4 + bx^3 + cx^2 + dx + e

The $g_2$ and $g_3$ invariants are

#stack,aligned g_2 = 60 \sum {\frac 1 {\omega^4}} = \tfrac {1} {12} (12ae - 3bd + c^2) g_3 = 140 \sum {\frac 1 {\omega^6}} = \tfrac {1} {432} (72ace - 27ad^2 - 27b^2e + 9bcd - 2c^3)

where the $\omega$ are the non-trivial periods of the integral of the holomorphic differential $dx/y$.

\omega = \bigint_C {1 \over \sqrt{ax^4 + bx^3 + cx^2 + dx + e}} dx

We want to carry out a similar computation for elliptic functions $f$ of order 3.


There is a special case answer to the above question:

The elliptic functions $f$ of order 3 with 3-fold rotational symmetry have differential equation

f'^3 = (af^3 + bf^2 + cf + d)^2

The associated curve of genus 1 is

y^3 = (ax^3 + bx^2 + cx + d)^2

The $g_2$ and $g_3$ invariants are

#stack,aligned g_2 = 60 \sum {\frac 1 {\omega^4}} = 0 g_3 = 140 \sum {\frac 1 {\omega^6}} = \tfrac {1} {729} (27a^2d^2 - 18abcd + 4ac^3 + 4b^3d - b^2c^2)

where, as above, the $\omega$ are the non-trivial periods of the integral of the differential $dx/y$.

\omega = \bigint_C {1 \over \sqrt[3]{(ax^3 + bx^2 + cx + d)^2}} dx

In both the above cases the $g_2$ and $g_3$ invariants are expressed in terms of the invariants of the curves coefficient polynomials. We seek something similar for the general order 3 elliptic function.


Editor's Note:

In the top right-hand corner of this page you can see the logo of a math typesetting library. Thanks to this excellent software I have been able to write up these web pages in a simple text editor in plain HTML and $\TeX$.

I have written these notes in an expository style. You will find details missing, a few over-simplifications, and many unsubstantiated assertions. But hopefully you will also find a coherent thread of reasoning from start to finish.

It has been a pleasure writing up these notes. Please enjoy!


Gregg Kelly

gregg.r.kelly@gmail.com