Remark 1 (Disclaimer) This article has been written for my own reference in preparation to my talk in occasion of the RNTA5 conference.
The distribution of the prime numbers is very chaotic and we do not have efficient rules to exaclty find the -th prime as a function of , whereas, thanks to the prime number theorem, we can certainly approximate the -th prime with a suitable function of (i.e., ). Indeed, most of the cryptography is based on the difficulty to factorize a number into its prime factors.
In particular, the distribution of gaps between primes behave like a random object, especially if we look at small scales, i.e. at gaps between consecutive primes smaller than the average gap (which by the prime number theorem is for ).
For this reason a well-studied problem was the following: fix a positive integer and a positive real number. How many short intervals of the form contain at least primes?
In his breakthrough paper on bounded gaps between primes, James Maynard proved that the answer is: infinitely many. In particular, if we set , we can be more specific and state that there exist infinitely many intervals of length at most containing at least primes.
Afterwards, Maynard improved his result, allowing a certain rate of uniformity, showing a more precise estimate:
if . The method used is a variation of a well-known sieve method, namely the GPY’s sieve.
These statements do not preclude the possibility that there are choices of and for which the intervals contain exactly primes, for at most finitely many . Anyway, this was proven to be not the case by Freiberg, who showed the following more precise result. For any positive real number and any non-negative integer , we have
if is sufficiently large in terms of and , and is a certain function that tends to zero as tends to infinity. We may take .
Since we are working with very short intervals we expect that an application of the well-known Cramér model for the distribution of prime numbers would give the exact answer for the distribution of such intervals. Therefore, heuristically we expect the above cardinality to be
for every and .
The idea behind Freiberg’s result is that we can make use of the Maynard’s sieve method to find clusters of consecutive primes inside particular sets and then construct short intervals of specific form around them. Indeed, the Maynard’s sieve method allows us to show that any subset of the primes, which is well distributed in arithmetic progressions (e.g., that verifies a variant of the prime number theorem and of the Bombieri-Vinogradov’s theorem), contains many elements that are close together. The work of Freiberg showed that the subset of primes which belongs to the image of certain admissible sets of linear functions, i.e.
is well distributed in arithmetic progressions and is suitable for the application of Maynard’s results. More specifically, when considering , we ask for a suitable very large in terms of and an admissible set with . We recall that a set is admissible if for any prime the set of all residue classes modulo do not cover all the classes mod . Next, the Maynard’s result tells us that the number of integers for which there are at least primes among is at least of order . A combinatorial process is hence used to detect exactly primes among them that are contained in our selected set of intervals. The idea indeed, is to start with a suitable interval of the form , which contains all the primes (at least ) inside . Next, we note that there exists a suitable such that and . Considering the sequence , it is easy to prove that for at least one we must have . Collecting the various intervals found in this way and using the estimates contained in the Maynard’s work, we recover the Freiberg’s result.
The major difference between the work of Freiberg and that one of the author is in the way the needed admissible set of linear forms is generated. In the former case, an Erdhos–Rankin type construction was considered, which allows us to lower bound the density related to each choice of and , choosing a suitable large value and an admissible set . However, this freedom inevitably forces us to lose precision and obtain weak estimates. In the latter case, the set of linear forms was chosen implicitly by means of the Maynard’s sieve, producing in this way better information on the density, allowing to consider of order , only for very small values of .
A better exploration of the last aforementioned approach, in which we insert a well-separated admissible set into the uniform Maynard’s estimates, leads us to generate a positive proportion of short intervals containing a prescribed number of primes. The key idea is that at the start of the process we need to select clusters of primes in which the elements are also well-spaced.
From now on, we will indicate with a non-negative integer, with the value , for certain suitable constants , and with a positive real number smaller than . The result is the following.
Theorem 1 (M.) We have
for a certain absolute constant , if is sufficiently large in terms of and .
2. Finding many intervals with clusters of well-spaced primes
We define the double sum
where equals to
Here means that the sum is over all the admissible sets such that and , for any , where a suitable constant. Moreover, we define
We consider verifying suitable growth conditions to be chosen later, a natural number under certain arithmetic conditions, and take as the sieve weights used in the Maynard’s work.
We can prove that
where now the set contains intervals of the form , for , with the property that , for a unique admissible set such that and , for any . We recall that the intervals in are pairwise disjoint, if for instance . The idea is to bound the term in parenthesis with times the indicator function that such term is positive, using the arithmetic definitions of the sums present there to define the type of intervals contained in .
Choosing sufficiently large in terms of , e.g. , , with , and and finally taking and suitably large, we may conclude that
for a certain constant .
with an absolute constant .
3. Combinatorial process
Consider an interval . There exist an integer and an admissible set , with and for any , such that and
In order to avoid having a trivial gap between the elements of we ask for to be sufficiently large with respect to and . Let us define
for . We recall here the following properties of the intervals :
1) for any such we have ;
2) for the choice we find that ;
3) for the value we have ;
4) if , for a certain , then
Now, let’s define
Note that we necessarily have being prime. Consequently, this implies , but from our assumption on it actually derives that
4. Final estimate
The information (10) is equivalent to say that we have found -different intervals containing exactly primes, with , if is sufficiently large. Together with the lower bound (8), we have obtained that for every and for each ,
which in turn is equivalent to
when is large enough in terms of and , for a certain constant .
5. Concluding remarks
The strength of the Maynard’s sieve is the flexibility, that makes it applicable to counting primes in sparser subsets as well. In fact, the same proof that leads to Theorem 1 can be overall adapted to study a variety of different situations, in which for instance we restrict the primes to lye on an arithmetic progression or allow for uniformity of the parameters and . The results are as following.
Remark 2 Let be sufficiently large in terms of and . Suppose that is a positive integer, with , as . Take with . Then, we have
for a certain , where is the intersection of with .
Remark 3 Fix a small parameter and . Let , and , obeying to the relations and . Then, the estimate (3) continues to hold.
Remark 4 Let be a Galois extension of with discriminant . There exist constants depending only on such that the following holds. Let be a conjugacy class in the Galois group of , and let
where denotes the Artin symbol. Let , and . Then, we have
If we consider values of slightly bigger than , a little variation of the sieve method used to prove Theorem 1 leads to the following improvement on the Freiberg bound.
Remark 5 For every non-negative integer and positive real number smaller than , with the value , for suitable constants , we have
for a certain , if is sufficiently large in terms of and .