Finite Fields (more)
William Stein
Jan 22, 2016
Motivation: if you plan to ever do any computations over finite fields, today is going to make your life easier later. Simple as that.
Update on the "saga" of GF(9)
.
Then Nathann Cohen complained, which led to a huge thread and this ticket: http://trac.sagemath.org/ticket/19929
some discussion
a rude/mean comment from Nathann directed at me...
ticket closed
another ticket: http://trac.sagemath.org/ticket/17569
It looks like doing GF(9)
might in the near future construct the quadratic subfield of . Is this a good idea? What is in Sage anyways?
Let's find out!
The lattice of finite fields...?
Bummer?
No, not really...
There's no reason to expect a+b to make any sense in Sage, since we have two finite fields above with two different variable. Also, if the names were the same, it also wouldn't make sense, because then would be the generator of and simultaneously.
Incidentally, Magma doesn't care what you call things, and does make sense of adding the and defined above:
Back to Sage...
At least we can do everything very explicitly and define embeddings.
So... at least it's possible to play around with different finite fields.
Though, as you might imagine, this "naive" appraoch can get very confusing and problematic! What if you do some more computations, choosing embeddings, and end up with two different ways to get from to , say? In a program it could happen very naturally, and lead to subtle bugs.
However, Sage does support working in , which is possibly very nice and solves our problem in a way that doesn't run into subtle bugs/contradictions if things get more complicated.
Exercise: Whenever you try anything in software you should be very worried that the implementation sucks. You should run some basic benchmarks and compare with what you know. Never trust anything (especially do not use closed source software). Is Fbar slow or fast? Try some simple benchmarks right now.
How does in Sage work?
The main idea is to use pseudo-conway polynomials, which are conway polynomials without the lexicographic condition.
Definition: Suppose is the minimal polynomial of a generator of . For , let be the smallest power of that generators , and let be the minimal polynomial of . Say that and are compatible of divides .
We call a pseudo-Conway polynomial if it is compatible with for all . This means that you can very easily understand the subfields of and maps between them explicitly in terms of these polynomials.
Of course finding a pseudo-Conway polynomial isn't trivial.
To make it even harder, you could try to find a Conway polynomial which is the same as above, except you require all the to be lexicographically minimal.
Exercise: Compare field creation time.
Next time (again delayed): finite fields from prime ideals in rings of integers of number fields