The famous Hardy-Littlewood conjecture on prime-tuples states that if $\{h_1, \cdots, h_k\} = \mathcal{H}$ is an admissible set, that is, for every prime $p$ the set $\mathcal{H}$ does not contain a complete residue system modulo $p$, then there exist infinitely many positive integers $n$ such that the tuple

$$\displaystyle (n+h_1, n+h_2, \cdots, n+h_k)$$

consists of only prime numbers. That is, each entry is prime.

The study of this conjecture is intimately connected to the study of bounded gaps between primes. Indeed, the confirmation that there exist infinitely often bounded gaps between primes is a confirmation of a weaker form of the above conjecture, namely that for fixed positive integer $k$ and a integer $l \geq 1$ (where we would like $l$ to be as small as possible the product

$$\displaystyle (n+h_1)\cdots(n+h_k)$$ contains at most $k+l$ many prime factors (counting multiplicity) between them, and for $l$ sufficiently small this in particular implies that there are at least two factors $n+h_i, n+h_j$ are prime.

My question concerns a higher dimensional analogue of the Hardy-Littlewood conjecture, namely looking at linear forms of more than one variable. Suppose that we are given the linear polynomials $L_j(x_1, \cdots, x_m) = a_1^{(j)} x_1 + \cdots + a_m^{(j)} x_m + h_j$, $1 \leq j \leq k$, such that $a_i^{(j)} \in \mathbb{N}$ for each $i,j$, and the content of each $L_j$ is 1. Then does one expect that for infinitely many $\textbf{x} = (x_1, \cdots, x_m) \in \mathbb{N}^m$ that the tuple

$$\displaystyle (L_1(\textbf{x}), \cdots, L_k(\textbf{x}))$$

is a prime tuple (that is, every entry is prime)?

I suspect that this would be much easier to prove than the Hardy-Littlewood conjecture due to the many degrees of freedom available when $m \geq 2$. If so, has anyone related results already?

  • $\begingroup$ For the record, all of this generalizes to multivariable polynomials of any degree (not just all linear), quantitatively and not just qualitatively. This would be a multivariable extension of the Bateman-Horn conjecture for single variable polynomials taking prime values. See Conjectures 2.2 and 2.3 of guests.mpim-bonn.mpg.de/kdestagnol/bathory.pdf. (Although that paper came out this year, in 2018, by no means is the conception of a multivariable Bateman-Horn conjecture that new. I think I have seen it in print, but at the moment am unable to find a definite earlier reference.) $\endgroup$
    – KConrad
    Oct 31, 2018 at 17:00

1 Answer 1


This is sometimes called Dickson's conjecture (questions of this form were raised in his 1904 paper, available here). Tao and Green's paper Linear Equations in the Primes give unconditional results along these lines, when combined with the resolution of the Gowers' norm inverse conjecture by Green, Tao and Ziegler, and the Mobius nilsequence conjecture by Green and Tao.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.