Short Intervals
The Prime Number Theorem describes the average distribution of primes up to a large number $x$:
Prime Distribution on Small Scales
The Prime Number Theorem describes the average distribution of primes up to a large number $x$:
$$ \pi(x)\sim \frac{x}{\log x}. $$
From this, one expects that an interval of length $h$ near $x$ should contain roughly
$$ \frac{h}{\log x} $$
primes.
Thus the quantity
$$ \pi(x+h)-\pi(x) $$
should approximately equal
$$ \frac{h}{\log x}. $$
The study of such questions is called the theory of primes in short intervals.
The main problem is determining how small the interval length $h$ may be while this approximation remains valid.
Average Prime Density
The heuristic density of primes near $x$ is
$$ \frac1{\log x}. $$
Therefore, integrating over a short interval gives the prediction
$$ \int_x^{x+h}\frac{dt}{\log t} \approx \frac{h}{\log x}, $$
provided $h$ is small compared with $x$.
This suggests
$$ \pi(x+h)-\pi(x) \sim \frac{h}{\log x}. $$
However, proving such estimates is difficult because prime numbers exhibit substantial local irregularity.
Trivial Limitations
If $h$ is extremely small, the approximation cannot hold uniformly.
For example, if
$$ h<2, $$
the interval may contain either zero or one prime. The predicted value
$$ \frac{h}{\log x} $$
is then too small to control the actual fluctuations.
Moreover, arbitrarily long gaps between consecutive primes exist. Thus one cannot expect every sufficiently short interval to contain a prime.
The challenge is finding the correct threshold for asymptotic behavior.
Intervals of Polynomial Length
A classical theorem states that if
$$ h=x^\theta $$
with
$$ \theta>1, $$
then the Prime Number Theorem immediately implies
$$ \pi(x+h)-\pi(x) \sim \frac{h}{\log x}. $$
The real difficulty begins when
$$ \theta<1. $$
Intervals shorter than $x$ probe the fine-scale structure of prime distribution.
Hoheisel’s Theorem
One of the first major advances was obtained by entity["people","Guido Hoheisel","German mathematician"] in 1930.
He proved that there exists a constant
$$ \theta<1 $$
such that
$$ \pi(x+x^\theta)-\pi(x) \sim \frac{x^\theta}{\log x}. $$
This result showed that primes occur regularly even inside intervals substantially shorter than $x$.
Later work gradually reduced the value of $\theta$.
Primes Between Consecutive Powers
A related question asks whether every interval
$$ [n^2,(n+1)^2] $$
contains a prime.
This is known as Legendre’s conjecture and remains open.
However, short-interval results imply weaker statements. For sufficiently large $x$, intervals of the form
$$ [x,x+x^\theta] $$
contain primes whenever $\theta$ exceeds certain explicit constants.
The best known unconditional exponents arise from deep estimates for zeros of the zeta function and exponential sums.
Chebyshev Functions in Short Intervals
The weighted function
$$ \psi(x) = \sum_{p^k\leq x}\log p $$
is often easier to analyze than $\pi(x)$.
The expected asymptotic relation becomes
$$ \psi(x+h)-\psi(x)\sim h. $$
This formulation connects directly with explicit formulas involving zeros of the zeta function.
Connection with the Riemann Hypothesis
The Riemann Hypothesis predicts strong control of primes in short intervals.
Assuming the hypothesis, one obtains
$$ \psi(x+h)-\psi(x) = h + O\left( \sqrt{x}(\log x)^2 \right). $$
Consequently, if
$$ h\gg \sqrt{x}(\log x)^2, $$
then the main term dominates the error, giving
$$ \pi(x+h)-\pi(x) \sim \frac{h}{\log x}. $$
Thus the Riemann Hypothesis predicts nearly optimal short-interval behavior.
Cramér’s Model
A probabilistic model introduced by entity["people","Harald Cramér","Swedish mathematician"] treats prime occurrence near $n$ as a random event with probability
$$ \frac1{\log n}. $$
This heuristic predicts that intervals of length roughly
$$ (\log x)^2 $$
should usually contain primes.
Although the model captures many statistical features correctly, rigorous proofs remain far beyond current methods.
Almost All Intervals
Even when uniform results are difficult, one can often prove statements for almost all intervals.
For example, many theorems show that for most $x$,
$$ \pi(x+h)-\pi(x) \sim \frac{h}{\log x} $$
holds for values of $h$ much smaller than those known in uniform estimates.
Such results use mean-square methods, zero-density estimates, and harmonic analysis.
Importance
Short intervals reveal the local structure of prime distribution. While the Prime Number Theorem describes global averages, short-interval theory investigates how evenly primes are spread on finer scales.
The subject connects deeply with:
- zeros of the zeta function,
- exponential sums,
- sieve methods,
- probabilistic models,
- additive combinatorics.
Many famous open problems, including prime gaps and conjectures about consecutive primes, belong naturally to the theory of short intervals.