2

without knowing any deeper theory, I am required to find the Weierstrass normal form of an elliptic curve, i.e. a representation of type $y^2z-x^3-axz-bz^3$ where $x,y,z $ are variables and $a,b$ are coefficients that have to be determined. The given curve is the vanishing set of $f=x^3+y^3+z^3$. My question: Is there a systematic way (beyond "guessing" coordinate transformations) to proceed in order to achieve this representation?

Thank you very much in advance!

JohnSmith
  • 1,514

1 Answers1

2

There are indeed algorithms to put an elliptic curve into Weierstrass normal form by prescribing a particular change of coordinates, e.g. Nagell's algorithm. (As Jan-Magnus Okland has pointed in the comments, many of these algorithms are implemented in Maple and other softwares.)

In the special case of the Fermat cubic $x^3 + y^3 + 1$, Nagell's algorithm tells us that we should make the substitution $x = \frac{36-v}{6u}$ and $y=\frac{36+v}{6u}$ to get the Weierstrass form $$ v^2 = u^3-432. $$

For many families of elliptic curves (e.g. the twisted Fermat cubics, Selmer curves, and Desboves curves, just to name a few), Nagell's algorithm and others have been worked out in detail to provide us with the general change of coordinates to put a member of this family into Weierstrass form (this is what was used above for the standard Fermat cubic).

msteve
  • 6,068
  • There might be a typo here: should the Fermat cubic be defined by $x^3 + y^3 = 1$? –  Oct 03 '22 at 02:36