A thin light curve on a dark ground: the path of zeta of one half plus it in the complex plane as t runs from 0 to 80, a tangle of loops that all pass through one point, the origin, once for every zero on the critical line
Unit 13 · Advanced + Analytic Number Theory · 75 min

The Riemann Zeta Function

Complex numbers, the zeta function and its continuation, its zeros, and the Riemann hypothesis
Image: Own rendering

The first unit of the analytic masterclass builds the complex numbers from scratch and follows Riemann's paper of 1859: the zeta function and its Euler product for complex s, the continuation past s = 1 by partial summation and the pole there, the functional equation with the trivial zeros and the values at 0 and at minus 1, the critical strip and the first zeros on the critical line, and the Riemann hypothesis. The explicit formula then writes the staircase of the prime powers as the line x minus one wave for every pair of zeros, the prime number theorem becomes the statement that no zero lies on the line of real part 1, and the unit ends with what is known: the proportion of zeros proved to lie on the critical line, the twelve trillion zeros verified by computer, the pair correlation of Montgomery and Dyson, and Robin's criterion.

Preview: the beginning of the unit

Unit 8 introduced the zeta function for real \(s > 1\) and Unit 12 used it to count primes, and both units said that the proof of the prime number theorem needs \(\zeta(s)\) for complex \(s\). This unit supplies the missing tool. It builds the complex numbers from scratch, extends \(\zeta\) to the whole plane except one point, finds its zeros, and shows in what sense the zeros are the primes seen from the other side: every zero contributes one wave to the staircase of the prime powers, and the prime number theorem is the statement that no zero lies on one particular line. The one claim in the unit that nobody has proved, the Riemann hypothesis, is the most famous open problem in mathematics, and the unit ends with what is known about it.

1 Complex numbers and complex powers

A complex number is an expression \(z = a + bi\) with real numbers \(a\) and \(b\) and a symbol \(i\) that obeys the one rule \(i^{2} = -1\). Complex numbers are added and multiplied like polynomials in \(i\), so \((a + bi)(c + di) = (ac - bd) + (ad + bc)i\), where the rule \(i^{2} = -1\) turns the term \(bdi^{2}\) into \(-bd\). The number \(a\) is the real part \(\Re z\) and \(b\) the imaginary part \(\Im z\). A complex number is a point \((a, b)\) of the plane, and two measurements of that point are used constantly. The modulus \(|z| = \sqrt{a^{2} + b^{2}}\) is its distance from the origin, and the argument \(\arg z\) is the angle, measured anticlockwise, from the positive real axis to the direction of \(z\). The complex conjugate \(\bar{z} = a - bi\) is the mirror image of \(z\) in the real axis, and \(z\bar{z} = a^{2} + b^{2} = |z|^{2}\). The modulus behaves like an absolute value: \(|zw| = |z|\,|w|\), and \(|z + w| \le |z| + |w|\), which is the triangle inequality.

The exponential function extends to complex arguments by Euler’s formula, which this unit uses as a definition: \begin{equation} e^{i\theta} = \cos\theta + i\sin\theta , \qquad e^{a + bi} = e^{a}\,(\cos b + i\sin b) . \label{eq:euler} \end{equation} The point \(e^{i\theta}\) lies on the circle of radius \(1\) at the angle \(\theta\), and the rule \(e^{z}e^{w} = e^{z + w}\) holds for all complex \(z\) and \(w\), because the addition formulas for sine and cosine are exactly what is needed to multiply two such points. Every complex number can therefore be written as \(z = |z|\,e^{i\arg z}\), and multiplying two numbers in this form multiplies their moduli and adds their angles, which is where the rule \(|zw| = |z|\,|w|\) comes from.

With the exponential in hand, a positive integer can be raised to a complex power. For a positive integer \(n\) and a complex number \(s\), which analytic number theory always writes as \[ s = \sigma + it , \qquad \sigma = \Re s , \quad t = \Im s , \] the power is defined through the natural logarithm of \(n\): \begin{equation} n^{s} = e^{s\ln n} = e^{\sigma\ln n}\,e^{it\ln n} = n^{\sigma}\bigl(\cos(t\ln n) + i\sin(t\ln n)\bigr) . \label{eq:npower} \end{equation} Two consequences of this definition are used throughout the unit. The modulus of \(n^{s}\) is \(n^{\sigma}\), because \(|e^{it\ln n}| = 1\), so the size of a complex power depends only on the real part of the exponent. The argument of \(n^{s}\) is \(t\ln n\), so the imaginary part of \(s\) only rotates the power. The rule \(n^{s}m^{s} = (nm)^{s}\) holds because \(\ln n + \ln m = \ln nm\), and \(n^{-s} = 1/n^{s}\) has modulus \(n^{-\sigma}\).