The Riemann hypothesis

The prime numbers are the ones you cannot split into a neat multiplication: 2, 3, 5, 7, 11, 13, and on. Every other whole number is built by multiplying primes together, so they are the building blocks of arithmetic.

Look at them up close and they seem to turn up wherever they like, with no pattern you can predict. There is no simple rule that tells you the next prime. But step back and look at millions of them at once and something surprising happens. They thin out in a smooth, dependable way. The bigger the numbers get, the rarer primes become, and they get rarer at a rate you can write down almost exactly.

So the primes are two things at the same time. Close up they look random. Far back they look like clockwork. Mathematicians even have a formula that estimates how many primes there are up to any number, and it works remarkably well.

The Riemann hypothesis is a single unproven guess about how good that estimate really is. It says the error, the gap between the estimate and the true count, is always as small as it could possibly be, never wandering off. More than 160 years later nobody has been able to prove it or find a case where it fails. It is one of the seven Millennium Prize Problems, and whoever settles it wins a million dollars.

Start with the counting. Write \(\pi(x)\) for the number of primes up to \(x\). The prime number theorem, proved in 1896, says that for large \(x\) this count is very close to \(x / \ln x\), and closer still to the logarithmic integral \(\operatorname{li}(x)\). The primes thin out like one in \(\ln x\): around a million, roughly one number in fourteen is prime. That is the smooth, large-scale law hiding behind the local chaos.

The tool that explains why is the Riemann zeta function. For a number \(s\) bigger than 1 it is an infinite sum,

\[ \zeta(s) = 1 + \frac{1}{2^{s}} + \frac{1}{3^{s}} + \frac{1}{4^{s}} + \cdots = \sum_{n=1}^{\infty} \frac{1}{n^{s}}. \]

Euler noticed this same sum can be rewritten as a product running over every prime, which is why zeta carries information about the primes at all. So far it only makes sense when the sum settles down, meaning the real part of \(s\) is greater than 1. Riemann's move was to extend it. There is exactly one sensible way to continue the function to almost every complex number \(s\), a process called analytic continuation. Think of it as the unique smooth surface that matches the original formula where they overlap and flows on naturally everywhere else.

Once you allow complex inputs, you can ask where zeta equals zero. Some zeros are easy and dull: at \(s = -2, -4, -6, \dots\) These are the trivial zeros. All the interesting ones live in the critical strip, the band where the real part of \(s\) is between 0 and 1. The Riemann hypothesis is the claim that every one of these non-trivial zeros has real part exactly \(\tfrac{1}{2}\). In other words, they all lie on a single vertical line, the critical line \(\operatorname{Re}(s) = \tfrac{1}{2}\).

Why would anyone care where an abstract function vanishes? Because those zeros control the error in the prime count. The smooth estimate \(\operatorname{li}(x)\) is the main term, and every zero adds a correcting ripple on top. If all the zeros sit on the critical line, those ripples stay tightly bounded and the estimate is as accurate as it can be. If even one zero drifted off the line, the primes would clump or thin in ways the smooth law could never track. The widget above shows the size of zeta straight up that line: every time the curve touches zero, you are looking at one of those zeros.

Zeros as the music of the primes. The link between the zeros and the primes is made exact by an explicit formula. Working with the prime-counting function (or the closely related Chebyshev function \(\psi(x) = \sum_{p^{k}\le x}\ln p\)), Riemann and von Mangoldt showed

\[ \psi(x) = x \;-\; \sum_{\rho} \frac{x^{\rho}}{\rho} \;-\; \ln(2\pi) \;-\; \tfrac{1}{2}\ln\!\big(1 - x^{-2}\big). \]

The leading \(x\) is the smooth prime number theorem. The sum runs over the non-trivial zeros \(\rho = \beta + i\gamma\), and each contributes an oscillating term. A zero at height \(\gamma\) produces a wave whose size grows like \(x^{\beta}\), where \(\beta\) is its real part. The zeros are, quite literally, the frequencies in the music of the primes: their heights set the pitches, their real parts set the volumes.

What the hypothesis buys you. If every \(\beta = \tfrac{1}{2}\), every oscillating term grows like \(x^{1/2}\), and the whole error collapses to

\[ \pi(x) = \operatorname{li}(x) + O\!\big(\sqrt{x}\,\ln x\big). \]

This is the smallest error the primes could possibly obey, and it is equivalent to the Riemann hypothesis: RH holds if and only if the prime number theorem carries this square-root error bound. A single zero with \(\beta > \tfrac{1}{2}\) would inflate the error to order \(x^{\beta}\) and let the primes drift much further from \(\operatorname{li}(x)\) than anyone observes.

The functional equation and the symmetry. Riemann proved that zeta obeys a reflection, tidily written with the completed function \(\xi(s) = \tfrac{1}{2}s(s-1)\pi^{-s/2}\Gamma(\tfrac{s}{2})\zeta(s)\), which satisfies \(\xi(s) = \xi(1 - s)\). This equation pairs the point \(s\) with \(1 - s\), so the non-trivial zeros are symmetric about the line \(\operatorname{Re}(s) = \tfrac{1}{2}\). The critical line is exactly the axis of that symmetry, which is what makes it the natural home for the zeros. The functional equation also produces the trivial zeros at the negative even integers, from the poles of the Gamma factor, cleanly separating them from the non-trivial ones inside the strip.

Where things stand. It is known that infinitely many zeros lie on the critical line, that a positive proportion of all zeros do, and that the first several trillion computed zeros are on it without exception. Zeros are also known to lie strictly inside the open strip \(0 < \operatorname{Re}(s) < 1\), which already yields the prime number theorem. But none of this rules out a stray zero off the line further up, and that gap is the whole difficulty. The Riemann hypothesis is one of the seven Millennium Prize Problems set by the Clay Mathematics Institute in 2000; six remain open, and this is widely regarded as the deepest of them.