Bill Allombert on Mon, 04 Dec 2023 00:49:31 +0100

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: correction to previous post

On Sun, Dec 03, 2023 at 02:59:52PM -0800, American Citizen wrote:
> In regard to heights of points across isogenous curves
> On 12/2/23 11:08, Bill Allombert wrote:
> > ? ellisomat(E)[2][,1]
> > %5 = [1,2,4,4,2,2,4,4]
> As stated this should be the same in Q as in a number field.
> I think that one of my pari programs was checking for saturation by a prime
> and when it used 2, it was able to find a smaller 1/2 height point on some
> of the isogenous curves, hence my relative vector did not match, but was
> altered from the normal result in %5.
> This begs the question, given all the curves in the ellisomat() command can
> we ferret out those who have a 2-degree function such that we can reliably
> locate where the situation occurs and the relative height vector is altered?

I am not sure I understand. On one hand, ellisomat(E)[2] return the degrees
of the isogenies between all the curve, on the other hand, by using
ellisdivisible and elltors (or over Q, ellsaturation) you can saturate the
points to get the generator.

> I have a follow up question, suppose we select a point on an elliptic curve
> in a number field, and in my case I picked a rational x-coordinate, and
> deriving y as a quadratic surd,  which gave me the number field to use. Are
> we able to find another independent point in the same number field also on
> the curve and I am NOT talking about doubling the given point and adding or
> subtracting multiples to give ht, 4ht, 9ht, 16ht etc, but a truly
> independent point in the same NF. I tried a naive approach and quickly
> realized the chances of hitting this was incredibly low.
> Or is this as hard as trying to find a rational point on a curve in Q?

Yes. If P is a point in a number fields, the sum of its conjugates (that you can
compute with elltrace) is a rational point. 
So the only easy points P are the ones which give a torsion point (or the point
at infinity).

In particular, if you pick x in Q and y in a quadratic extension, you will get
the point at infinity (this is easy to prove).