Working notes on zeta zeros
These pages are Grant Golland’s working notes on the zeros of the Riemann zeta function: computed figures, cited theorems, and the project’s own proofs and computations, each flagged. The Riemann Hypothesis, posed by Bernhard Riemann in 1859, is one open conjecture about these zeros. Nothing here proves it.
Every figure is computed here or taken from a cited table. Sentences carry a flag: Settled, Open, or Speculative.
How these notes were made
Method and status
The label “Proof'd”
The instructions field of the Grok project is two words: Status: Proof'd. Grant clarified, for this site, that the label was an aspirational lock on the conceptual architecture. It was not a serious claim that the Riemann Hypothesis had been proved. On this site it always sits next to the plain statement that nothing here proves the Riemann Hypothesis.
Clay Millennium Prize
A project conversation answers a direct question about applying for the prize. The answer recorded there is no.
- Settled Institute rules The Clay Mathematics Institute does not accept direct submissions. A proposed solution is considered only after publication in a qualifying outlet, at least two years from publication, and general acceptance in the global mathematics community. The rules are published by the Institute: Rules for the Millennium Prize Problems.
Figures, and what they are not
On 2 October 2026 PT the site gained three computed figures: the first five nontrivial zeros on the critical line, a five-pair truncation of the explicit formula for ψ(x), and π(x) against li(x) up to 200. Checkpoints live with the research notes. A figure that uses five known zeros is not a proof about the sixth. On 3 October 2026 PT the count was extended to twenty zeros, π(x) against li(x) out to 1000 (π(1000) = 168), and the explicit-formula truncation to twenty pairs. The longer truncation is numerically stable and the residual does not shrink at every x; that is reported, not smoothed over. |ζ(1/2+it)| for t ≤ 50 was plotted from computed values. Twenty zeros on the line are not the Riemann Hypothesis. On 4 October 2026 PT the signed error li(x) − π(x) was plotted to 10 000; every sample stayed positive, with max |li−π| ≈ 23.336 at x = 8622 on the grid. No sign change is claimed. On 6 October 2026 PT four more computed figures were added to the ζ page: Hardy’s Z(t) on [0, 60], whose 13 sign changes reproduce the first 13 ordinates; the zero count N(T) to T = 200 against the Riemann–von Mangoldt main term (N(200) = 79, smooth term 79.193); a nearest-neighbour spacing histogram of the first 100 000 zeros from Odlyzko’s published table, set against the GUE Wigner surmise and Poisson; and Riemann’s R(x) against li(x) to 10 000 (max |R − π| ≈ 5.41, max |li − π| ≈ 23.34). The spacing histogram supports a conjecture about how zeros repel. It is not a test of whether they lie on the line. A second round the same day added three more figures from Odlyzko’s tables. First, pair correlation of the first 100 000 zeros against Montgomery’s 1 − (sin πu/πu)², with root-mean-square deviation 0.028. Second, nearest-neighbour spacings for 10 000 zeros near the 1012-th and 1022-th zero against the exact GUE law: variance 0.177 and 0.176, against 0.180 for GUE and 0.161 for the low zeros. Third, S(T) up to the 100 000th zero, whose mean square sits a small, nearly constant amount above Selberg’s (1/2π²) log log T. All three concern conjectured statistics or classical theorems on finite ranges. None bears on whether every zero is on the line. A third round added two more computed figures and the site’s first interactive one. Number variance, the variance of the count of zeros in a window of L mean spacings, follows GUE closely near the 1022-th zero (0.541 at L = 10, GUE 0.579) and levels off early for the low zeros (0.339), in the order Berry’s 1988 heuristic predicts. Gram’s law, one zero per Gram interval, holds in 79.4% of intervals low down, 70.6% near the 1012-th zero and 69.1% near the 1022-th. The interactive figure lets the reader add zeros to the explicit formula one pair at a time; the average residual falls (root mean square 1.571 with none, 0.499 with 100 pairs) while the residual beside prime powers barely moves. These are statistics and arithmetic on known zeros, not steps toward a proof.
Reading path · step 11 of 13