Riemann Zeta Zeros — Critical Line Visualizer
Plots the Riemann–Siegel Z(t) function (real, zero-crossings on Re(s)=1/2 ≡ zeros of ζ(s)) and the trace of ζ(1/2+it) in the complex plane.
Scroll to zoom, drag to pan, double-click a chart to reset it.
Z(t) vs t — zero crossings = zeta zeros
ζ(1/2 + it) traced in the complex plane
Proportion of zeros proven on the critical line, by mathematician
Detected zeros (t where Z(t) = 0)
Methods, visually
Each proof pushes the bound with a different tool. The formulas below are the actual objects used — not illustrations of them.
1914 · Hardy
Studied Z(t) itself (the real function plotted above) and showed it changes sign infinitely often on t ∈ (0, T) — each sign change is a zero on the critical line. The chart below is literally Hardy's argument: crossings that never stop.
1942 · Selberg
First proof that a positive proportion (not just an infinite count) of zeros lie on the line, via mean-value estimates of ζ. The original proof gives no explicit numeric c — just that one exists.
1974 · Levinson
The mollifier: a short Dirichlet polynomial that, multiplied against ζ, cancels its growth and straightens the spiral above — turning zero-detection into counting sign changes of a simpler function. This is the "mollifier length (y)" slider above.
1989 · Conrey
Same mollifier idea as Levinson, pushed further: a longer, higher degree Dirichlet polynomial combined with sharper 4th-moment estimates for ζ. More terms → tighter cancellation → more provable zeros.
2020 · Pratt, Robles, Zaharescu, Zeindler
Layers a new zero-density input on top of the mollifier framework: zeros are provably sparser away from Re(s)=1/2, which is what lets a mollifier of bounded length catch a larger share of them. (Density sketch, not measured data.)
2026 · Claude — More Than Two Thirds…
A different tool than the mollifier line above: Montgomery's 1973 pair-correlation sum, unconditionally. Off-line zero pairs {ρ, 1−ρ̄} contribute hyperbolic 2×2 blocks (signature (1,1)) to a Gram matrix; on-line simple zeros contribute clean positive-definite 1×1 blocks. Sylvester's law of inertia makes that split basis-independent, and a rank–trace inequality (proved via von Neumann's trace inequality — the matrix analogue of Montgomery's real-number step m² ≥ 2m−1) bounds how much of the matrix the off-line pairs can occupy. RHS is what used to need RH to read; here it's linear algebra instead.
Preprint, not yet peer-reviewed — the strongest claim in this README, but still a claim.