In number theory, a formula for primes is a formula generating the prime numbers, exactly and without exception. Formulas for calculating primes do exist; however, they are computationally very slow. A number of constraints are known, showing what such a "formula" can and cannot be.

A simple formula is

${\displaystyle f(n)=\left\lfloor {\frac {n!{\bmod {(}}n+1)}{n}}\right\rfloor (n-1)+2}$

for positive integer ${\displaystyle n}$, where ${\displaystyle \lfloor \ \rfloor }$ is the floor function, which rounds down to the nearest integer. By Wilson's theorem, ${\displaystyle n+1}$ is prime if and only if ${\displaystyle n!\equiv n{\pmod {n+1}}}$. Thus, when ${\displaystyle n+1}$ is prime, the first factor in the product becomes one, and the formula produces the prime number ${\displaystyle n+1}$. But when ${\displaystyle n+1}$ is not prime, the first factor becomes zero and the formula produces the prime number 2.^[1] This formula is not an efficient way to generate prime numbers because evaluating ${\displaystyle n!{\bmod {(}}n+1)}$ requires about ${\displaystyle n-1}$ multiplications and reductions modulo ${\displaystyle n+1}$.

In 1964, Willans gave the formula

${\displaystyle p_{n}=1+\sum _{i=1}^{2^{n}}\left\lfloor \left({\frac {n}{\sum _{j=1}^{i}\left\lfloor \left(\cos {\frac {(j-1)!+1}{j}}\pi \right)^{2}\right\rfloor }}\right)^{1/n}\right\rfloor }$

for the ${\displaystyle n}$th prime number ${\displaystyle p_{n}}$.^[2] This formula reduces to^[3][4] ${\displaystyle p_{n}=1+\sum _{i=1}^{2^{n}}[\pi (i)<n]}$; that is, it tautologically defines ${\displaystyle p_{n}}$ as the smallest integer m for which the prime-counting function ${\displaystyle \pi (m)}$ is at least n. This formula is also not efficient. In addition to the appearance of ${\displaystyle (j-1)!}$, it computes ${\displaystyle p_{n}}$ by adding up ${\displaystyle p_{n}}$ copies of ${\displaystyle 1}$; for example, ${\displaystyle p_{5}=1+1+1+1+1+1+1+1+1+1+1+0+0+\dots +0=11}$.

The articles What is an Answer? by Herbert Wilf (1982)^[5] and Formulas for Primes by Underwood Dudley (1983)^[6] have further discussion about the worthlessness of such formulas.

Because the set of primes is a computably enumerable set, by Matiyasevich's theorem, it can be obtained from a system of Diophantine equations. Jones et al. (1976) found an explicit set of 14 Diophantine equations in 26 variables, such that a given number k + 2 is prime if and only if that system has a solution in nonnegative integers:^[7]

${\displaystyle \alpha _{0}=wz+h+j-q=0}$

${\displaystyle \alpha _{1}=(gk+2g+k+1)(h+j)+h-z=0}$

${\displaystyle \alpha _{2}=16(k+1)^{3}(k+2)(n+1)^{2}+1-f^{2}=0}$

${\displaystyle \alpha _{3}=2n+p+q+z-e=0}$

${\displaystyle \alpha _{4}=e^{3}(e+2)(a+1)^{2}+1-o^{2}=0}$

${\displaystyle \alpha _{5}=(a^{2}-1)y^{2}+1-x^{2}=0}$

${\displaystyle \alpha _{6}=16r^{2}y^{4}(a^{2}-1)+1-u^{2}=0}$

${\displaystyle \alpha _{7}=n+\ell +v-y=0}$

${\displaystyle \alpha _{8}=(a^{2}-1)\ell ^{2}+1-m^{2}=0}$

${\displaystyle \alpha _{9}=ai+k+1-\ell -i=0}$

${\displaystyle \alpha _{10}=((a+u^{2}(u^{2}-a))^{2}-1)(n+4dy)^{2}+1-(x+cu)^{2}=0}$

${\displaystyle \alpha _{11}=p+\ell (a-n-1)+b(2an+2a-n^{2}-2n-2)-m=0}$

${\displaystyle \alpha _{12}=q+y(a-p-1)+s(2ap+2a-p^{2}-2p-2)-x=0}$

${\displaystyle \alpha _{13}=z+p\ell (a-p)+t(2ap-p^{2}-1)-pm=0}$

The 14 equations α₀, …, α₁₃ can be used to produce a prime-generating polynomial inequality in 26 variables:

${\displaystyle (k+2)(1-\alpha _{0}^{2}-\alpha _{1}^{2}-\cdots -\alpha _{13}^{2})>0.}$

That is,

${\displaystyle {\begin{aligned}&(k+2)(1-{}\\[6pt]&[wz+h+j-q]^{2}-{}\\[6pt]&[(gk+2g+k+1)(h+j)+h-z]^{2}-{}\\[6pt]&[16(k+1)^{3}(k+2)(n+1)^{2}+1-f^{2}]^{2}-{}\\[6pt]&[2n+p+q+z-e]^{2}-{}\\[6pt]&[e^{3}(e+2)(a+1)^{2}+1-o^{2}]^{2}-{}\\[6pt]&[(a^{2}-1)y^{2}+1-x^{2}]^{2}-{}\\[6pt]&[16r^{2}y^{4}(a^{2}-1)+1-u^{2}]^{2}-{}\\[6pt]&[n+\ell +v-y]^{2}-{}\\[6pt]&[(a^{2}-1)\ell ^{2}+1-m^{2}]^{2}-{}\\[6pt]&[ai+k+1-\ell -i]^{2}-{}\\[6pt]&[((a+u^{2}(u^{2}-a))^{2}-1)(n+4dy)^{2}+1-(x+cu)^{2}]^{2}-{}\\[6pt]&[p+\ell (a-n-1)+b(2an+2a-n^{2}-2n-2)-m]^{2}-{}\\[6pt]&[q+y(a-p-1)+s(2ap+2a-p^{2}-2p-2)-x]^{2}-{}\\[6pt]&[z+p\ell (a-p)+t(2ap-p^{2}-1)-pm]^{2})\\[6pt]&>0\end{aligned}}}$

is a polynomial inequality in 26 variables, and the set of prime numbers is identical to the set of positive values taken on by the left-hand side as the variables a, b, …, z range over the nonnegative integers.

A general theorem of Matiyasevich says that if a set is defined by a system of Diophantine equations, it can also be defined by a system of Diophantine equations in only 9 variables.^[8] Hence, there is a prime-generating polynomial inequality as above with only 10 variables. However, its degree is large (in the order of 10⁴⁵). On the other hand, there also exists such a set of equations of degree only 4, but in 58 variables.^[9]

The first such formula known was established by W. H. Mills (1947), who proved that there exists a real number A such that, if

${\displaystyle d_{n}=A^{3^{n}}}$

then

${\displaystyle \left\lfloor d_{n}\right\rfloor =\left\lfloor A^{3^{n}}\right\rfloor }$

is a prime number for all positive integers n.^[10] If the Riemann hypothesis is true, then the smallest such A has a value of around 1.3063778838630806904686144926... (sequence A051021 in the OEIS) and is known as Mills' constant.^[11] This value gives rise to the primes ${\displaystyle \left\lfloor d_{1}\right\rfloor =2}$, ${\displaystyle \left\lfloor d_{2}\right\rfloor =11}$, ${\displaystyle \left\lfloor d_{3}\right\rfloor =1361}$, ... (sequence A051254 in the OEIS). Very little is known about the constant A (not even whether it is rational). This formula has no practical value, because there is no known way of calculating the constant without finding primes in the first place.

There is nothing special about the floor function in the formula. Tóth proved that there also exists a constant ${\displaystyle B}$ such that

${\displaystyle \lceil B^{r^{n}}\rceil }$

is also prime-representing for ${\displaystyle r>2.106\ldots }$.^[12]

In the case ${\displaystyle r=3}$, the value of the constant ${\displaystyle B}$ begins with 1.24055470525201424067... The first few primes generated are:

${\displaystyle 2,7,337,38272739,56062005704198360319209,}$

${\displaystyle 176199995814327287356671209104585864397055039072110696028654438846269,\ldots }$

Without assuming the Riemann hypothesis, Elsholtz developed several prime-representing functions similar to those of Mills. For example, if ${\displaystyle A=1.00536773279814724017\ldots }$, then ${\displaystyle \left\lfloor A^{10^{10n}}\right\rfloor }$ is prime for all positive integers ${\displaystyle n}$. Similarly, if ${\displaystyle A=3.8249998073439146171615551375\ldots }$, then ${\displaystyle \left\lfloor A^{3^{13n}}\right\rfloor }$ is prime for all positive integers ${\displaystyle n}$.^[13]

Another tetrationally growing prime-generating formula similar to Mills' comes from a theorem of E. M. Wright. He proved that there exists a real number α such that, if

${\displaystyle g_{0}=\alpha }$ and

${\displaystyle g_{n+1}=2^{g_{n}}}$ for ${\displaystyle n\geq 0}$,

then

${\displaystyle \left\lfloor g_{n}\right\rfloor =\left\lfloor 2^{\dots ^{2^{2^{\alpha }}}}\right\rfloor }$

is prime for all ${\displaystyle n\geq 1}$.^[14] Wright gives the first seven decimal places of such a constant: ${\displaystyle \alpha =1.9287800}$. This value gives rise to the primes ${\displaystyle \left\lfloor g_{1}\right\rfloor =\left\lfloor 2^{\alpha }\right\rfloor =3}$, ${\displaystyle \left\lfloor g_{2}\right\rfloor =13}$, and ${\displaystyle \left\lfloor g_{3}\right\rfloor =16381}$. ${\displaystyle \left\lfloor g_{4}\right\rfloor }$ is even, and so is not prime. However, with ${\displaystyle \alpha =1.9287800+8.2843\cdot 10^{-4933}}$, ${\displaystyle \left\lfloor g_{1}\right\rfloor }$, ${\displaystyle \left\lfloor g_{2}\right\rfloor }$, and ${\displaystyle \left\lfloor g_{3}\right\rfloor }$ are unchanged, while ${\displaystyle \left\lfloor g_{4}\right\rfloor }$ is a prime with 4932 digits.^[15] This sequence of primes cannot be extended beyond ${\displaystyle \left\lfloor g_{4}\right\rfloor }$ without knowing more digits of ${\displaystyle \alpha }$. Like Mills' formula, and for the same reasons, Wright's formula cannot be used to find primes.

Given the constant ${\displaystyle f_{1}=2.920050977316\ldots }$ (sequence A249270 in the OEIS), for ${\displaystyle n\geq 2}$, define the sequence

${\displaystyle f_{n}=\left\lfloor f_{n-1}\right\rfloor (f_{n-1}-\left\lfloor f_{n-1}\right\rfloor +1)}$

(1)

where ${\displaystyle \left\lfloor \ \right\rfloor }$ is the floor function. Then for ${\displaystyle n\geq 1}$, ${\displaystyle \left\lfloor f_{n}\right\rfloor }$ equals the ${\displaystyle n}$th prime: ${\displaystyle \left\lfloor f_{1}\right\rfloor =2}$, ${\displaystyle \left\lfloor f_{2}\right\rfloor =3}$, ${\displaystyle \left\lfloor f_{3}\right\rfloor =5}$, etc. ^[16] The initial constant ${\displaystyle f_{1}=2.920050977316}$ given in the article is precise enough for equation (1) to generate the primes through 37, the ${\displaystyle 12}$th prime.

The exact value of ${\displaystyle f_{1}}$ that generates all primes is given by the rapidly-converging series

${\displaystyle f_{1}=\sum _{n=1}^{\infty }{\frac {p_{n}-1}{P_{n}}}={\frac {2-1}{1}}+{\frac {3-1}{2}}+{\frac {5-1}{2\cdot 3}}+{\frac {7-1}{2\cdot 3\cdot 5}}+\cdots ,}$

where ${\displaystyle p_{n}}$ is the ${\displaystyle n}$th prime, and ${\displaystyle P_{n}}$ is the product of all primes less than ${\displaystyle p_{n}}$. The more digits of ${\displaystyle f_{1}}$ that we know, the more primes equation (1) will generate. For example, we can use 25 terms in the series, using the 25 primes less than 100, to calculate the following more precise approximation:

${\displaystyle f_{1}\simeq 2.920050977316134712092562917112019.}$

This has enough digits for equation (1) to yield again the 25 primes less than 100.

As with Mills' formula and Wright's formula above, in order to generate a longer list of primes, we need to start by knowing more digits of the initial constant, ${\displaystyle f_{1}}$, which in this case requires a longer list of primes in its calculation.

In 2018 Simon Plouffe conjectured a set of formulas for primes. Similarly to the formula of Mills, they are of the form

${\displaystyle \left\{a_{0}^{r^{n}}\right\}}$

where ${\displaystyle \{\ \}}$ is the function rounding to the nearest integer. For example, with ${\displaystyle a_{0}\approx 43.80468771580293481}$ and ${\displaystyle r=5/4}$, this gives 113, 367, 1607, 10177, 102217... (sequence A323176 in the OEIS). Using ${\displaystyle a_{0}=10^{500}+961+\varepsilon }$ and ${\displaystyle r=1.01}$ with ${\displaystyle \varepsilon }$ a certain number between 0 and one half, Plouffe found that he could generate a sequence of 50 probable primes (with high probability of being prime). Presumably there exists an ε such that this formula will give an infinite sequence of actual prime numbers. The number of digits starts at 501 and increases by about 1% each time.^[17][18]

It is known that no non-constant polynomial function P(n) with integer coefficients exists that evaluates to a prime number for all integers n. The proof is as follows: suppose that such a polynomial existed. Then P(1) would evaluate to a prime p, so ${\displaystyle P(1)\equiv 0{\pmod {p}}}$. But for any integer k, ${\displaystyle P(1+kp)\equiv 0{\pmod {p}}}$ also, so ${\displaystyle P(1+kp)}$ cannot also be prime (as it would be divisible by p) unless it were p itself. But the only way ${\displaystyle P(1+kp)=P(1)=p}$ for all k is if the polynomial function is constant. The same reasoning shows an even stronger result: no non-constant polynomial function P(n) exists that evaluates to a prime number for almost all integers n.

Euler first noticed (in 1772) that the quadratic polynomial

${\displaystyle P(n)=n^{2}+n+41}$

is prime for the 40 integers n = 0, 1, 2, ..., 39, with corresponding primes 41, 43, 47, 53, 61, 71, ..., 1601. The differences between the terms are 2, 4, 6, 8, 10... For n = 40, it produces a square number, 1681, which is equal to 41 × 41, the smallest composite number for this formula for n ≥ 0. If 41 divides n, it divides P(n) too. Furthermore, since P(n) can be written as n(n + 1) + 41, if 41 divides n + 1 instead, it also divides P(n). The phenomenon is related to the Ulam spiral, which is also implicitly quadratic, and the class number; this polynomial is related to the Heegner number ${\displaystyle 163=4\cdot 41-1}$. There are analogous polynomials for ${\displaystyle p=2,3,5,11{\text{ and }}17}$ (the lucky numbers of Euler), corresponding to other Heegner numbers.

Given a positive integer S, there may be infinitely many c such that the expression n² + n + c is always coprime to S. The integer c may be negative, in which case there is a delay before primes are produced.

It is known, based on Dirichlet's theorem on arithmetic progressions, that linear polynomial functions ${\displaystyle L(n)=an+b}$ produce infinitely many primes as long as a and b are relatively prime (though no such function will assume prime values for all values of n). Moreover, the Green–Tao theorem says that for any k there exists a pair of a and b, with the property that ${\displaystyle L(n)=an+b}$ is prime for any n from 0 through k − 1. However, as of 2020,^[update] the best known result of such type is for k = 27:

${\displaystyle 224584605939537911+18135696597948930n}$

is prime for all n from 0 through 26.^[19] It is not even known whether there exists a univariate polynomial of degree at least 2, that assumes an infinite number of values that are prime; see Bunyakovsky conjecture.

Another prime generator is defined by the recurrence relation

${\displaystyle a_{n}=a_{n-1}+\gcd(n,a_{n-1}),\quad a_{1}=7,}$

where gcd(x, y) denotes the greatest common divisor of x and y. The sequence of differences a_n+1 − a_n starts with 1, 1, 1, 5, 3, 1, 1, 1, 1, 11, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 23, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 47, 3, 1, 5, 3, ... (sequence A132199 in the OEIS). Rowland (2008) proved that this sequence contains only ones and prime numbers. However, it does not contain all the prime numbers, since the terms gcd(n + 1, a_n) are always odd and so never equal to 2. 587 is the smallest prime (other than 2) not appearing in the first 10,000 outcomes that are different from 1. Nevertheless, in the same paper it was conjectured to contain all odd primes, even though it is rather inefficient.^[20]

Note that there is a trivial program that enumerates all and only the prime numbers, as well as more efficient ones, so such recurrence relations are more a matter of curiosity than of any practical use.

• Prime number theorem

• Regimbal, Stephen (1975), "An explicit Formula for the k-th prime number", Mathematics Magazine, 48 (4), Mathematical Association of America: 230–232, doi:10.2307/2690354, JSTOR 2690354.
• A Venugopalan. Formula for primes, twinprimes, number of primes and number of twinprimes. Proceedings of the Indian Academy of Sciences—Mathematical Sciences, Vol. 92, No 1, September 1983, pp. 49–52 errata

• Weisstein, Eric W., "Prime Formulas" ("Prime-Generating Polynomial") at MathWorld.