The Millennium Lab

Millennium Problems

Seven rounds of real computation against six open problems. Nothing solved. Every number traced to a run, every run to a receipt.

Program
The Millennium Lab
Rounds
7
Receipts
66, chain valid, head bdebd1e7
Date
Nov 20-21, 2026
Images
14
Status
NOT SOLVED
Problems
6
Sorries
0 (Hodge)
NOT SOLVED. Nothing on this site solves the problem, proves it, or disproves it. This is research: measurements, rejections, and partial formalizations, each with a receipt. The problem belongs to the mathematicians who stated it and the researchers who have worked it for decades. This program adds a small computational exploration on top of their work and claims no credit for theirs.

01The six problems, and what this program did about them

Seven rounds, November 20-21, 2026. 66 receipts, hash-chained, head bdebd1e7d93dd108bffe0ba01d499e603db81a8c17027bb0aad36a70f73dacdd. Nothing is solved. Each problem below has its own page with everything in it: the statement, the method, the round-by-round numbers, the figures, the receipts, the limits, and the next attack.

Riemann Hypothesis

First 4,000 zeros re-verified on the critical line, two independent implementations, max real-part deviation 0.0 at 30 digits. Top of range t = 4506.31. Re-verification, not a new record. Read the page (rounds 1, 2, 4, 5). Poster: millennium-riemann-poster.png.

Birch and Swinnerton-Dyer

130 elliptic curves, conductor to 50: 128 rank 0 with L(E, 1) nonzero, 2 rank 1 (37a1, 43a1) with L(E, 1) = 0 and L'(E, 1) nonzero: 0.306000 and 0.343524. The pattern holds on every curve checked. Read the page (rounds 1, 2, 4, 5). Poster: millennium-bsd-poster.png.

P versus NP

One concrete decision problem from a real codebase. Registry search in P (0.0718 s). The defeater-free fragment in P by greedy search (44 of 44 against brute force). The full fragment with defeaters also in P (8,876 checks, 0 violations). The hardness direction stalls on structure, and existence is 1,400 of 1,400 at every density. Greedy extends to 5 of 6 non-composer ops. Read the page (rounds 1-5). Poster: millennium-p-vs-np-poster.png.

Hodge conjecture

Formalized in Lean 4.13.0: 58 theorems checked, 0 sorries. Round 5 alone: 18 theorems, the sheaf long exact sequence with c1 for the actual exponential sequence. The Lefschetz (1,1) case is now conditional on exactly two named inputs (Dolbeault isomorphism, Hodge decomposition). Read the page (rounds 1-5). Poster: millennium-hodge-poster.png.

Navier-Stokes

Full 96^3 at nu = 0.000125: Kida-Pelz enstrophy growth 145.79, Taylor-Green 138.27, both smooth, no blowup, no saturation. Growth is transient, not runaway: at 64^3 both flows peak and decay by T = 20. The viscosity ladder at 64^3 is monotone: 18.86 / 40.33 / 64.46 / 88.24. Read the lab (rounds 1-7). Field renders below, r1 to r7.

Yang-Mills

The torelon channel is overturned. Three seeds disagree at beta 2.5 (1.2798 / 0.8899 / 1.0795) and at beta 2.6 (1.6084 / 1.1931 / 1.2675), and the estimator itself has no stable plateau: the mass moves more across tau (2.395 at beta 2.5, 2.100 at beta 2.6) than across chains at fixed tau (0.390, 0.415). No continuum extrapolation is attempted. Read the lab (rounds 1-7).

02Gallery: all 14 images

Every image is 1080x1350 and computed from the lab's own solvers or data. Field renders: panel 1 is a Kida-Pelz vorticity slice (z = pi plane) at peak enstrophy; panel 2 is a Yang-Mills SU(2) plaquette weave (xy plane, z = 0, t = 0). Rounds 3-4 onward panel 2 is drawn twice, plain and after 4 APE smearing steps (alpha 0.5).

millennium-field-render.png
millennium-field-render.png The original first render. Round 1. Navier-Stokes 48^3, nu 0.001, enstrophy growth 18.29, peak t = 7.8; Yang-Mills beta 2.3, plaquette mean 0.536.
millennium-field-render-r1.png
millennium-field-render-r1.png Round 1. Same Navier-Stokes run; Yang-Mills beta 2.2.
millennium-field-render-r2.png
millennium-field-render-r2.png Round 2. Navier-Stokes 64^3, nu 0.001, growth 18.86, peak t = 7.7; Yang-Mills beta 2.3.
millennium-field-render-r3.png
millennium-field-render-r3.png Round 3. Navier-Stokes 64^3 run to t = 20; Yang-Mills beta 2.5.
millennium-field-render-r34.png
millennium-field-render-r34.png Rounds 3-4. Navier-Stokes 64^3, nu 0.0005, growth 40.33, peak t = 7.8; Yang-Mills beta 2.3, plain and after 4 APE smearing steps.
millennium-field-render-r4.png
millennium-field-render-r4.png Round 4. Finalized version of the r34 field, same numbers.
millennium-field-render-r5.png
millennium-field-render-r5.png Round 5. Navier-Stokes 64^3, nu 0.00025, growth 64.46, peak t = 8.9; Yang-Mills beta 2.6.
millennium-field-render-r6.png
millennium-field-render-r6.png Round 6. Navier-Stokes 64^3, nu 0.000125, growth 88.24, peak t = 9.4; Yang-Mills beta 2.5.
millennium-field-render-r7.png
millennium-field-render-r7.png Round 7. Navier-Stokes 96^3, nu 0.000125, growth 145.79, peak t = 8.2, snapshot taken from the hunt run itself; Yang-Mills beta 2.6.
millennium-round2-poster.png
millennium-round2-poster.png Round 2 poster, different layout, built by build_poster.py.
millennium-bsd-poster.png
millennium-bsd-poster.png BSD poster. 130 curves to conductor 50, rank split 128 rank 0 / 2 rank 1 (37a1, 43a1), L' values 0.306000 and 0.343524.
millennium-riemann-poster.png
millennium-riemann-poster.png Riemann poster. First 10 zeros on the critical line with exact t-values; counts 1,000 / 2,000 / 3,000 / 4,000.
millennium-p-vs-np-poster.png
millennium-p-vs-np-poster.png P vs NP poster. Phase transition vs p^r, dichotomy 8,876 checks / 0 violations, existence 1,400 / 1,400, greedy 5 of 6 ops.
millennium-hodge-poster.png
millennium-hodge-poster.png Hodge poster. New theorems per round 4 / 9 / 13 / 14 / 18 (0 sorries), exponential sequence and long exact sequence, both blocked inputs marked.

03Breakthroughs: what was resolved, and what was not

Nothing solved. Ten sub-questions were resolved along the way, each with its round and receipt. The full record, with data sources and the audit, is breakthroughs.html.

  1. Hodge: the sheaf long exact sequence is proved for the actual exponential sequence (round 5, rcpt_0057).
  2. Hodge: the structure sheaf O_X is tractable, overturning the round 2 assessment (round 3, rcpt_0047).
  3. Hodge: degree-zero Dolbeault, as far as mathlib allows, 14 new theorems, 0 sorries (round 4, rcpt_0051).
  4. P vs NP: the defeater-free fragment is POLYNOMIAL; the dichotomy holds, 8,876 checks, 0 violations (rounds 2-3, rcpt_0039, rcpt_0046).
  5. P vs NP: the p^r phase transition does NOT survive defeaters; existence is 1,400 of 1,400 at every density (round 4, rcpt_0050).
  6. P vs NP: greedy extends to 5 of 6 non-composer ops; CONTRADICTION_FREE is the exception (round 5, rcpt_0056).
  7. Riemann: the first 4,000 zeros are verified on the critical line (rounds 1, 2, 4, 5; rcpt_0001, rcpt_0042, rcpt_0052, rcpt_0058).
  8. BSD: analytic ranks computed for all 130 curves to conductor 50 (rounds 1, 2, 4, 5; rcpt_0002-0029, rcpt_0042, rcpt_0052, rcpt_0058).
  9. Navier-Stokes: enstrophy growth is transient, not runaway, at every tested viscosity (rounds 6-7, rcpt_0060, rcpt_0063).
  10. Navier-Stokes: 96^3 runs are now feasible, about 0.48 hours (rounds 6-7, rcpt_0060, rcpt_0063).

Still open: all six problems; the Hodge Dolbeault isomorphism and Hodge decomposition; a usable Yang-Mills channel; Navier-Stokes blowup at lower viscosity or higher resolution; P vs NP separation or collapse; BSD for all curves and Riemann for all zeros.

Overturned: the Yang-Mills torelon channel; the five-point continuum extrapolation; the p^r phase transition as a property of the full fragment; the 96^3 cost wall; the round 2 assessment that O_X is not tractable; anti-parallel vortex tubes as a blowup attack.

04Everything in one place

ItemWhere
This page: the six problems, the gallery, the breakthroughsindex.html
The lab, all seven roundslab.html
Claim landscape, all six problemsatlas.html
Problems indexproblems.html
Riemann pageriemann.html
Birch and Swinnerton-Dyer pagebsd.html
P versus NP pagep-vs-np.html
Hodge pagehodge.html
What was resolved, in fullbreakthroughs.html
All evidence: data, Lean, scripts, render sourcesresearch.html
Field renders and posters, rounds 1-7renders/ (14 images, 1080x1350)
The Millennium Lab. Research pages, November 2026. All numbers trace to the lab data files; every receipt is listed on the page. Nothing solved.