This interactive needs WebGL2 with floating-point render targets, which this browser does not provide.
AI Mathematics · Number theory
Riemann’s Line, 1859 → 2026 Phase · The zeta phase portrait
Every point of this plane is a number, and the zeta function sends it to another. The colour tells you which way the answer points. Where all the colours meet, the answer is zero.
What you are looking at Each point is a complex number s = σ + it. The real part σ runs across, the height t runs up. The zeta function turns s into another complex number, ζ(s), and every pixel is coloured by the direction of ζ(s) in the complex plane. The thin dark lines are contours of size: crossing one, |ζ(s)| has doubled or halved. Where there is room, a fainter line marks each quarter step.
The key is the same colouring applied to s itself: each direction from the centre has its colour, and each ring is one doubling of size. Pointing right means a positive real number, left a negative one. Nothing here is pre-drawn. ζ is computed afresh for each pixel of your screen as you move, on your graphics card.
What a zero is Where ζ(s) = 0 it has no direction, so every colour meets at one point, in order, and the contour lines close in around it in shrinking rings. At s = 1, where ζ grows without bound, the colours also meet, but turn the other way round. Along the real line to the left, at −2, −4, −6 and so on, lie the trivial zeros, which have been understood since Riemann.
Curiosities Places worth a closer look. Choose one to fly there; turn on Curiosities in the controls, or press c , to see their letters on the picture. The same place keeps its letter in all three views.
Along the middle line The path of ζ along the middle line, over the heights in your view: a curve that loops through the origin, the cyan point, exactly at each zero. The newest stretch, at the top of your view, is the brightest.
The critical strip All the other zeros lie in the strip between real parts 0 and 1, marked by two faint lines. The bright line through its middle is real part ½. The first zero sits at height 14.13; they crowd closer as you climb, about ln(t/2π)/2π of them per unit of height, so near height 10,000 they arrive every 0.86 or so. Odlyzko’s table lists the first 10,000, all on the line, up to height 9,878.
The Riemann Hypothesis Riemann conjectured in 1859 that every zero in the strip lies exactly on the middle line. The zeros control how irregularly the primes are spaced: the hypothesis is equivalent to the statement that the number of primes below x never strays from its smooth estimate by much more than √x. Fly up the line and watch each colour wheel land on it.
Zeros and primes The primes’ irregular spacing is captured by a smoothed prime sum, the one Allahyar’s reconstruction of the argument starts from. Each zero contributes one wave to it. The blue curve is built only from the primes; the gold one only from the waves of the zeros below the top of your view, so it closes in on the blue as you climb. Hover a zero on the line to draw its single wave in cyan.
G(x) = Σ_(n≤x) Λ(n)(x/n − 1)
− x log x + (1 + γ)x
Λ(pᵏ) = log p, otherwise 0
G(x) = log 2π − T(x) + Σ_ρ x^ρ/(ρ(1 − ρ))
T(x) = Σ x^(−2n)/(2n(2n + 1))
drawn = (G(x) − log 2π + T(x))/√x
= Σ_γ 2cos(γ log x)/(¼ + γ²)
if every ρ = ½ + iγ
band = Σ_ρ 1/(ρ(1 − ρ))
= 2 + γ − log 4π ≈ 0.0462
If every zero lies on the line, each wave keeps a fixed height forever, and the curve can never leave the cyan band. A single zero off the line, at real part β greater than ½, would add a wave growing like x to the power β − ½, and the curve would eventually break out. Computed from the primes up to two million, it never reaches more than about half the band’s height (0.0227, near x = 319,439).
The announcement On 20 November 2026, Levent Alpöge announced that, working with Claude, he had found a proof of the Riemann Hypothesis, extending to the Generalized Riemann Hypothesis for Dirichlet L-functions (his note ). He released it only as a formalization in the Lean proof assistant, 175,187 files and about 17.4 million lines, and wrote that a readable account would follow. The same day, Sahibzada Allahyar posted an expository reconstruction of the argument. It reduces the hypothesis to an upper bound on the prime sum drawn above, over windows from Y to Y^(3/2), obtained from a chain of about √x theta lifts.
Status This page does not check the proof. As of 21 November 2026 the claim is unverified: an independent build on a public server whose operator is not named reported on 20 November that the whole project compiles with no failures and that its final theorems rest only on Lean’s three standard axioms (a fresh kernel replay was skipped; the audit ), but no named mathematician or peer review has confirmed it; the reconstruction states that the full dependency tree and the axioms of the final declaration have not been independently checked. What the page shows is classical and independent of the claim. The zeros it can mark are numerical values from Odlyzko’s tables , and the hypothesis has been verified numerically for every zero up to height 3 × 10¹² (Platt and Trudgian, 2021). The explicit formula and the band are standard consequences of the zeros’ positions, checked here against the primes to two million; the gold curve uses only the zeros below the top of your view, so it is a truncated sum.
How it is computed Right of the middle line, ζ is summed by Euler–Maclaurin with about t/(1.4π) + 10 terms and sixteen Bernoulli corrections; to the left, the functional equation ζ(s) = χ(s) ζ(1 − s) is applied in logarithms, so the picture never overflows. The graphics card works in single precision, so the large phases t·ln n are prepared in double precision for each band of rows. Checked against a double-precision reference, the direction of ζ is right to within 0.03° away from the immediate neighbourhood of a zero. Heights to 15,000.
ζ(s) = Σ_(n<N) n^−s + N^(1−s)/(s−1) + ½N^−s + Σ_(j≤16) B₂ⱼ/(2j)! · s(s+1)…(s+2j−2) · N^(−s−2j+1)
colour = arg ζ(s) lines: log₂|ζ(s)| ∈ ℤ (and ¼ℤ)