Prime-counting function

For a real number x, π(x) denotes the number of primes p satisfying p ≤ x. The symbol π here names a function; it is not the circle constant. For example, the primes no greater than 10 are 2, 3, 5 and 7, so π(10) = 4. Since the count changes only when x reaches a prime, π(x) is a step function: it remains constant between successive primes and increases by one at each prime.

The prime number theorem describes the broad growth of π(x). In asymptotic notation, π(x) ~ x / ln(x) as x grows without bound; equivalently, the ratio of π(x) to x / ln(x) tends to 1. Thus x / ln(x) becomes a relatively accurate estimate of the count for sufficiently large x, even though it is not an exact formula. The theorem expresses the gradual thinning of prime numbers among larger integers: near a large value x, the rough density of primes is about 1 / ln(x).

In the late eighteenth century, Adrien-Marie Legendre proposed an approximation for the number of primes, and Carl Friedrich Gauss independently developed an insight into their distribution. Gauss later related prime counting to the logarithmic integral, a generally sharper approximation than x / ln(x). The prime number theorem was eventually proved independently in 1896 by Jacques Hadamard and Charles-Jean de la Vallée Poussin. Their proof established a precise limiting relationship, rather than merely a numerical pattern suggested by tables.

The function is central to analytic number theory because it turns the question “how many primes are there up to x?” into an object that can be studied and estimated. The prime number theorem gives large-scale average behaviour, not the location of each next prime or an exact count at a particular input. For any finite x, π(x) can in principle be found by counting primes up to x; estimates are useful when direct counting is impractical. More refined approximations, including those involving the logarithmic integral, can improve on the leading estimate in many ranges.