Answers are grounded in this paper and cite their sources.
We study the lag-one covariance of consecutive prime gaps g_n = p_(n+1) - p_n in the window [10^10, 3x10^10), after local detrending and conditioning on the opening-prime residue modulo a primorial M. For each of seven moduli M in {6, 30, 210, 2310, 30030, 510510, 9699690} we obtain an exact decomposition A + X + Y + D = Cov that closes to within 10^-6. The residual correlation rho_resid remains negative at all seven scales, decreasing in magnitude from -0.0247 to -0.0103. The between-phase variance grows monotonically with M, accounting for up to 3.74% of Var(e). The non-D terms account for 60.4% of the lag-one covariance at the largest tested modulus, while between-phase structure accounts for 3.74% of the one-point variance. We report block-bootstrap 95% intervals for rho_resid that exclude zero at every scale. The results are finite-range and do not claim asymptotic laws. Source code, CSV results, run log and SHA256SUMS checksums are attached to this record.
No takes yet. Share an insight, caveat, or question.
Reddouane BERRAMDANE (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: