Riemann ζ · change of variable s = w + c

Zeta Shift Lab

Slide the input by c and watch the grid fold under ζ. At c = ½ the critical line lands on the imaginary axis, so the line of symmetry sits at 0. Every value here is computed live in your browser and checked against published numbers below.

View
Function
0.50
Input w Output F(w)
Pointer w
move over the plane
s = w + c
–
F(w)
–
|F(w)|
–

Lines and regions

Positions here are in s, the original variable, so a region stays on the same part of ζ when you move the shift. On the plane it appears moved left by c (now 0.50). Leave a box empty for no limit.

Show
Main grid spacing
Finer grid in a region
Re s to Im s to every

    Your own lines
    for Im s to

      Walking up the critical line

      Hardy's Z(t) = eiθ(t) ζ(½ + it) has the same size as ζ on the critical line but is real, so each zero on the line shows up as a crossing of the axis. With c = ½ this is the function along the imaginary axis w = it. Red dots mark the zeros the page found.

      Checking the output

      G(w) = ξ(½ + w) is the completed function in your shifted coordinates. The symmetry rows compute both sides straight from the definition, without using the functional equation, so agreement is a real test.

      CheckComputedExpectedErrorResult
      Running…

      The shift relabels the plane

      Writing s = w + c moves every point by the same amount. Zeros, the pole and the critical line all slide together, and the folded picture keeps its shape. That is why the zeros still land on 0 for any c: try dragging the slider while the morph sits at the output end.

      Why ζ alone is lopsided

      The symmetry is between s and 1 − s, a mirror at Re s = ½. The line Re s = 1 is where the sum 1 + 2−s + 3−s + … stops converging, and s = 1 is the pole. ζ itself carries extra gamma factors, so it is not mirror-symmetric. The completed ξ(s) = ½ s(s−1) π−s/2 Γ(s/2) ζ(s) is. Switch to ξ in the phase portrait at c = ½ to see it mirror left-right and top-bottom.

      Riemann made the same move in 1859

      His paper works with Ξ(t) = ξ(½ + it). In this page's coordinates that is G(it): shift by ½, then rotate a quarter turn. G is even and real on the imaginary axis, and the Riemann Hypothesis becomes the statement that every zero of Ξ is real.

      Where current research sits

      The de Bruijn–Newman constant Λ measures how far a heat-flow deformation of Ξ can go before its zeros leave the real line. RH is equivalent to Λ ≤ 0. Rodgers and Tao proved Λ ≥ 0 in 2018, and the best upper bound is Λ ≤ 0.2 (Platt and Trudgian, 2021). If RH is true, it holds with no room to spare.

      Shifting the output is different

      Subtracting 1 moves where things land instead of relabelling where they start. The points that now map to 0 are the solutions of ζ(s) = 1, which are not the Riemann zeros. Tick the box and watch the red dots stop at −1.

      Method: Borwein's accelerated alternating series for Re s ≥ ½, the functional equation for Re s < ½, Lanczos log-gamma, double precision. Accurate to about 10−13 for |Im s| ≤ 350; the plane is not computed above that height. Drag to pan; pinch, Ctrl + scroll or the + and − buttons to zoom.