Independently recomputed constants (mpmath dps=40)
All values below were recomputed tonight, not copied from secondary blogs.
Hexagonal packing (target)
- Center density
δ_hex = 1/(2√3) = √3/6 = 0.2886751345948128822545743902509787278238 - Area density
Δ_hex = π/(2√3) = 0.906899682117108915361219637392477477... - Theorem 3.2 last-sign-change target:
2π r_*^2 = 4π/√3 = 7.255197456936871402376313030568622929136
Published Cohn–Elkies 2003 (Annals 157, 689–714)
- Table 3, n=2: best packing
0.28868, new upper bound0.28868(center density, five decimals). - Table 4, n=2:
2π r^2 = 7.25520, forced double roots21.77, 29.02, 50.79, 65.34, 90.19. - Recomputed from the printed Table 4
R:δ ≤ 7.25520/(8π) = 0.2886752357800797600176013700050665498581- ratio vs hex
= 1.000000350516046419403644002640063972807
Levenshtein / CE Proposition 6.1 (exact Bessel)
j_1 =first positive zero ofJ_1= 3.831705970207512315614435886308160766565Δ ≤ j_1^2/16 = 0.9176231651327433285762361105393816868551δ ≤ j_1^2/(16π) = 0.2920885252530132815247208990820543499602- ratio vs hex
= 1.011824332092168126325936766864430803816
This certificate (certs/ce_d2_m5.json)
R = 3627599/500000 = 7.255198δ ≤ R/(8π) = 0.2886751562026082140699334855631848636009- ratio vs hex
= 1.000000074851598708507619246976814816322 - Isolated last odd root of
G:(1267758233/174737932, 878736008/121118135)≈ (7.2551976464961250, 7.2551976464961259) - Meets CE Table 3 (
δprints as0.28868). - Strictly below CE Table 4 printed
R = 7.25520. - Does not attain
4π/√3. Not a magic function.