In number theory, a prime k-tuple is a finite collection of values representing a repeatable pattern of differences between prime numbers. For a k-tuple (a, b, …), the positions where the k-tuple matches a pattern in the prime numbers are given by the set of integers n such that all of the values (n + a, n + b, …) are prime. Typically the first value in the k-tuple is 0 and the rest are distinct positive even numbers.^[1]

Several of the shortest k-tuples are known by other common names:

OEIS sequence OEIS: A257124 covers 7-tuples (prime septuplets) and contains an overview of related sequences, e.g. the three sequences corresponding to the three admissible 8-tuples (prime octuplets), and the union of all 8-tuples. The first term in these sequences corresponds to the first prime in the smallest prime constellation shown below.

In order for a k-tuple to have infinitely many positions at which all of its values are prime, there cannot exist a prime p such that the tuple includes every different possible value modulo p. For, if such a prime p existed, then no matter which value of n was chosen, one of the values formed by adding n to the tuple would be divisible by p, so there could only be finitely many prime placements (only those including p itself). For example, the numbers in a k-tuple cannot take on all three values 0, 1, and 2 modulo 3; otherwise the resulting numbers would always include a multiple of 3 and therefore could not all be prime unless one of the numbers is 3 itself. A k-tuple that satisfies this condition (i.e. it does not have a p for which it covers all the different values modulo p) is called admissible.

It is conjectured that every admissible k-tuple matches infinitely many positions in the sequence of prime numbers. However, there is no admissible tuple for which this has been proven except the 1-tuple (0). In that case, the conjecture is equivalent to the statement that there are infinitely many primes. Nevertheless, Yitang Zhang proved in 2013 that there exists at least one 2-tuple which matches infinitely many positions; subsequent work showed that such a 2-tuple exists with values differing by 246 or less that matches infinitely many positions.^[2]

Although (0, 2, 4) is not admissible it does produce the single set of primes, (3, 5, 7).

Some inadmissible k-tuples have more than one all-prime solution. This cannot happen for a k-tuple that includes all values modulo 3, so to have this property a k-tuple must cover all values modulo a larger prime, implying that there are at least five numbers in the tuple. The shortest inadmissible tuple with more than one solution is the 5-tuple (0, 2, 8, 14, 26), which has two solutions: (3, 5, 11, 17, 29) and (5, 7, 13, 19, 31), where all values mod 5 are included in both cases.

The diameter of a k-tuple is the difference of its largest and smallest elements. An admissible prime k-tuple with the smallest possible diameter d (among all admissible k-tuples) is a prime constellation. For all n ≥ k this will always produce consecutive primes.^[3] (Recall that all n are integers for which the values (n + a, n + b, …) are prime.)

This means that, for large n:

${\displaystyle p_{n+k-1}-p_{n}\geq d}$

where p_n is the nth prime number.

The first few prime constellations are:

The diameter d as a function of k is sequence A008407 in the OEIS.

A prime constellation is sometimes referred to as a prime k-tuplet, but some authors reserve that term for instances that are not part of longer k-tuplets.

The first Hardy–Littlewood conjecture predicts that the asymptotic frequency of any prime constellation can be calculated. While the conjecture is unproven it is considered likely to be true. If that is the case, it implies that the second Hardy–Littlewood conjecture, in contrast, is false.

A prime k-tuple of the form (0, n, 2n, 3n, …, (k − 1)n) is said to be a prime arithmetic progression. In order for such a k-tuple to meet the admissibility test, n must be a multiple of the primorial of k.^[5]

The Skewes numbers for prime k-tuples are an extension of the definition of Skewes' number to prime k-tuples based on the first Hardy–Littlewood conjecture (Tóth (2019)). Let ${\displaystyle P=(p,\ p+i_{1},\ p+i_{2},\ \dots \ ,\ p+i_{k})}$ denote a prime k-tuple, ${\displaystyle \pi _{P}(x)}$ the number of primes p below x such that ${\displaystyle p,\ p+i_{1},\ p+i_{2},\ \dots \ ,\ p+i_{k}}$ are all prime, let ${\textstyle \operatorname {li} _{P}(x)=\int _{2}^{x}{\frac {dt}{(\ln t)^{k+1}}}}$ and let ${\displaystyle C_{P}}$ denote its Hardy–Littlewood constant (see first Hardy–Littlewood conjecture). Then the first prime p that violates the Hardy–Littlewood inequality for the k-tuple P, i.e., such that

${\displaystyle \pi _{P}(p)>C_{P}\operatorname {li} _{P}(p),}$

(if such a prime exists) is the Skewes number for P.

The table below shows the currently known Skewes numbers for prime k-tuples:

Prime k-tuple	Skewes number	Found by
⁠${\displaystyle (p,\ p+2)}$⁠	1369391	Wolf (2011)
⁠${\displaystyle (p,\ p+4)}$⁠	5206837	Tóth (2019)
⁠${\displaystyle (p,\ p+2,\ p+6)}$⁠	87613571	Tóth (2019)
⁠${\displaystyle (p,\ p+4,\ p+6)}$⁠	337867	Tóth (2019)
⁠${\displaystyle (p,\ p+2,\ p+6,\ p+8)}$⁠	1172531	Tóth (2019)
⁠${\displaystyle (p,\ p+4,\ p+6,\ p+10)}$⁠	827929093	Tóth (2019)
⁠${\displaystyle (p,\ p+2,\ p+6,\ p+8,\ p+12)}$⁠	21432401	Tóth (2019)
⁠${\displaystyle (p,\ p+4,\ p+6,\ p+10,\ p+12)}$⁠	216646267	Tóth (2019)
⁠${\displaystyle (p,\ p+4,\ p+6,\ p+10,\ p+12,\ p+16)}$⁠	251331775687	Tóth (2019)
⁠${\displaystyle (p,\ p+2,\ p+6,\ p+8,\ p+12,\ p+18,\ p+20)}$⁠	7572964186421	Pfoertner (2020)
⁠${\displaystyle (p,\ p+2,\ p+8,\ p+12,\ p+14,\ p+18,\ p+20)}$⁠	214159878489239	Pfoertner (2020)
⁠${\displaystyle (p,\ p+2,\ p+6,\ p+8,\ p+12,\ p+18,\ p+20,\ p+26)}$⁠	1203255673037261	Pfoertner / Luhn (2021)
⁠${\displaystyle (p,\ p+2,\ p+6,\ p+12,\ p+14,\ p+20,\ p+24,\ p+26)}$⁠	523250002674163757	Luhn / Pfoertner (2021)
⁠${\displaystyle (p,\ p+6,\ p+8,\ p+14,\ p+18,\ p+20,\ p+24,\ p+26)}$⁠	750247439134737983	Pfoertner / Luhn (2021)

The Skewes number (if it exists) for sexy primes ⁠${\displaystyle (p,\;p+6)}$⁠ is still unknown.