Primer · before the zeros
Complex numbers: the plane behind the zeros
The zeros of ζ live in the complex plane. This page builds that plane from scratch: what i is, why anyone wanted it, how to picture it, and why the line with real part 1/2 matters. Every figure is computed, not drawn by hand.
1. A definition, not a discovery
No ordinary number squares to −1: a positive times itself is positive, and so is a negative. So mathematicians simply defined a new number, written i, by the rule
i² = −1.
Everything else follows by the usual rules of arithmetic. A complex number is anything of the form a + bi, where a and b are ordinary (real) numbers: a is its real part and b its imaginary part. The word “imaginary” is an accident of history; i is no less legitimate than −1 or √2.
2. Why anyone wanted it: cubics, not quadratics
For a quadratic like x² + 1 = 0, people were happy to say “no solution.” The pressure came from cubic equations. Gerolamo Cardano published a formula for them in Ars Magna (1545). Rafael Bombelli, in his Algebra (1572), noticed that for x³ = 15x + 4 the formula asks for √−121, even though the equation has three perfectly real answers, one of them 4. He pushed through anyway, treating √−1 by fixed rules, and the imaginary parts cancelled to give the real answer.
Figure 1 Why i was invented: Bombelli’s cubic Settled
3. The picture: a second number line, at right angles
Where does i go on the number line? Nowhere: it gets its own line. Draw the real numbers left to right as usual, and the multiples of i (i, 2i, −i …) bottom to top. The two lines cross at 0 and together make a flat grid, the complex plane. The number 2 + 3i is the point 2 steps across and 3 steps up.
Settled A common misreading: i does not bend or curve the number line. The real line stays straight and the plane is flat. What i does is turn, as the next section shows. Caspar Wessel (1799), Jean-Robert Argand (1806) and Carl Friedrich Gauss (1831) made this picture standard; Gauss’s 1831 paper is why it is often called the Gaussian plane.
Figure 2 The complex plane: a second line at right angles Settled
4. Multiplying is turning and stretching
Every point of the plane has a length (its distance from 0) and an angle (measured anticlockwise from the positive real direction). The key fact, the reason complex numbers are useful, is:
to multiply two complex numbers, multiply their lengths and add their angles.
The number i has length 1 and angle 90°, so multiplying by i is a pure quarter turn: 1 → i → −1 → −i → 1. Two quarter turns make a half turn, which sends 1 to −1. That is all i² = −1 says.
Figure 3 Multiplying turns and stretches Settled
5. Euler’s formula: motion on a circle
Leonhard Euler, in his Introductio in analysin infinitorum (1748), found that the exponential function and the circle are the same thing seen two ways:
eit = cos t + i sin t.
Read it as motion: as t increases, the point eit walks around the circle of radius 1 at steady speed, having travelled a distance t along it. At t = π it has gone halfway round and sits at −1. That gives Euler’s identity, eiπ = −1 (often written eiπ + 1 = 0). Its honest meaning is not mystical: half a turn around the unit circle lands on −1. It is beautiful because it ties together the exponential, the circle (π) and turning (i).
Figure 4 Euler’s formula on the unit circle Settled
6. A function moves the whole plane
An ordinary function takes a number on a line to a number on a line. A complex function takes a point of the plane to a point of the plane, so the way to picture one is to draw a grid and see where every line goes. Squaring, z ↦ z², doubles each angle and squares each length: the grid bends into parabolas, still crossing at right angles. The exponential, z ↦ ez, wraps a strip around the origin, turning vertical lines into circles and horizontal lines into rays. The plane stays flat; it is the grid lines that the function bends.
Colouring each point by the angle of the output gives a phase portrait. For z² the colours go round the origin twice. The same colouring applied to ζ is the phase map of ζ, where each zero shows up as a point where all the colours meet.
Figure 5 A grid before and after z² and e^z Settled
7. The bridge to ζ
Now the zeros. ζ is a function from the plane to the plane, and a zero is an input s where ζ(s) = 0.
- Settled The imaginary axis is the vertical line where the real part is 0: numbers 0 + ti. The critical line is a different vertical line, half a unit to its right: numbers 1/2 + ti.
- Settled The critical strip is everything with real part strictly between 0 and 1. All the nontrivial zeros lie in it. The trivial zeros are −2, −4, −6, … on the real axis; they come from the functional equation and are well understood (see the trivial zeros).
- Settled Feed every point 1/2 + it of the critical line into ζ and plot the outputs: you get a curve that loops around the origin. Each time it passes exactly through 0 is a zero on the line. The first is at t ≈ 14.1347.
- Open The Riemann Hypothesis says every nontrivial zero has real part exactly 1/2. It is not proved. Every zero checked so far, up to height 3·1012, is on the line, but checking is not proof.
Figure 6 The critical line and the loop of ζ(1/2 + it) Settled
zetazero(k) for k = 1 to 10, and ζ(1/2 + it) at 1001 points with step 0.05 for 0 ≤ t ≤ 50 (values beyond ±4.2 clipped to the frame). ζ(1/2) ≈ −1.4604. The heights of the zeros on the left are the moments the right-hand curve hits 0. Settled as computation; the claim that all zeros behave this way is Open. Motion plays on load and is off if your system asks for reduced motion.8. A fun true fact: knight moves are Gaussian integers
Complex numbers a + bi with whole-number a and b are called Gaussian integers (Gauss studied them in 1831). Their norm is a² + b², the squared length. Put a knight on the square 0 of a board drawn on the plane: its eight jumps land exactly on the Gaussian integers of norm 5, namely ±1 ± 2i and ±2 ± i. That is true, but it is a curiosity, not a deep link between chess and ζ.
- Settled 5 = (1 + 2i)(1 − 2i). An odd prime splits into two Gaussian factors exactly when it leaves remainder 1 on division by 4 (Fermat’s two-squares theorem). Primes like 3 and 7, with remainder 3, do not split, so no leaper move has squared length 3 or 7. The race between these two kinds of prime is on the prime races page of the Prime Distribution site.
- Settled The number of Gaussian integers of norm n is 4 × (divisors of n that are 1 mod 4 − divisors that are 3 mod 4). The zeta function that counts them, the Dedekind zeta function of the Gaussian integers, factors as ζ(s) · L(s, χ4), where L(s, χ4) is the L-function of the remainder-mod-4 pattern. Its first zero is at 1/2 + 6.0209i (see zeros of other L-functions).
- Open Whether all nontrivial zeros of L(s, χ4) lie on the critical line is part of the Generalized Riemann Hypothesis. Results that depend on it are conditional.
Figure 7 Knight moves as Gaussian integers Settled
Where to go next
With the plane in hand, the next page, zeros on the critical line, follows ζ along the line 1/2 + it in detail.
Reading path · step 2 of 13