Given a integer , consider the set . For how many values of prime , the set exhausts all the non-zero residues. That is for how many primes , the given integer primitive element modulo ?
There are some obvious examples where there are a very few/no such primes. For instance take only generates one residue class and primitive only for . generates at most two residues classes and is primitive only for .
If is a perfect square, the residues generated will all be square residues and cannot exhaust Similar logic applies to perfect powers, but other powers residues can sometimes exhaust the whole group – for instance if
So the squares and seem to be the only issues and we have the following conjecture:
Conjecture. If or a perfect square, then there is a constant such that
For ,
That is the set
of primes with primitive element has density among all the primes.
Below we will discuss a conditional proof under GRH. The problem is still open, but the state of the art is Heath-Brown proof that at least one of 2, 3, or 5 is a primitive root modulo infinitely many primes
Hooley’s Proof conditional on GRH:
Let us fix . Now is a primitive root modulo iff
for . That is we dont want
to hold for any prime
We first estimate the set of primes for which there are no obstructions like above for . We also estimate , the primes for which there is an obstruction , then our required count for primes where 2 is primitive is given because those elements of that shouldn’t be included in our count are all elements of
By inclusion-exclusion we write as sifted sum where we remove primes for which the obstruction holds at different primes
Now each of the term inside can be seen as the number of splitting primes in the compositum of Kummer extensions for . So by GRH (for the Dedekind zeta function of this compositum), we get the count split primes
Plugging this in we get
by taking and observing that
To estimate the primes that have a obstruction , we split the range of primes into three parts small, medium and large (, , and
Again for each of the primes , these counts are exactly split primes in the Kummer extension .
For small primes, summing over all the gives a bound .
For medium primes we just use the condition and estimate the primes in arithmetic progressions by Brun-Titchmarsh
gives an admissible bound like
Finally for the large primes, use the fact that , to bound the number of primes by the number of prime factors of all the numbers , where
The number of prime factors of is small by simple estimates like .
Putting all these together we get the desired estimate
C. Hooley, On Artin’s conjecture, J. reine angew. Math. 225 (1967), 209-220.
D. R. Heath-Brown, Artin’s conjecture for primitive roots, Quart. J. Math. Oxford Ser. (2) {3 7}(1986), 27-38.
One thought on “Artin Primitive Roots Conjecture (Hooley’s Conditional Proof)”