> David Roe and I started looking into this at SD3. What David Harvey
> said in his reply is right -- there's no way to choose a canonical
> map in the mathematical sense, but you can make a "default" choice. I
> know David Roe has thought more about this, especially wrt the coerce
> conversations
> * How to pick a canonical isomorphism? My idea is to pick the roots
> that
> define phi and phi^(-1) (r1, r2) treat them as polynomials over
> GF(q).base_ring() and choose the canonical isomorphism where the
> product
> r1*r2 is minimal. Does this make sense?
>
> * From there it should be ea
On Jun 26, 2007, at 8:39 AM, Martin Albrecht wrote:
>
> Hi there,
>
> I have given some thought to finite extension field isomorphisms
> and finite
> extension field embeddings.
Before you get too deeply into this, you should also discuss this
with David Roe (and possibly others), who has b
