Збережено в:
| Автор: | |
|---|---|
| Формат: | Recurso digital |
| Мова: | Англійська |
| Опубліковано: |
Zenodo
2026
|
| Предмети: | |
| Онлайн доступ: | https://doi.org/10.5281/zenodo.18181737 |
| Теги: |
Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
|
Зміст:
- <p>We report a geometric point pattern model that treats Earth as a sphere and asks whether a small set of globally distributed "anchor sites" can be connected by a fixed menu of target distances defined as fractions of a single global scale, λ. The model is intentionally agnostic about any ancient-use story: it outputs coordinates; independent Earth-system layers are then used to test whether those coordinates are measurably non-random. Using a fixed parameter set (λ≈38,287.67 km; fractions F={1/20, 1/5, 1/3}; tolerance ε=1%), we observe: (1) a statistically significant network among 12 anchor sites vs. latitude-preserving null worlds (p<0.0001 at target scale); (2) out-of-sample prediction-held-out anchors land closer to model-generated candidates than expected by chance (75.5% win-rate, p<10■¹²); (3) enrichment for proximity to oil/gas deposits and metallic mineral deposits at the anchor coordinates (BH-FDR corrected); (4) elevated karst-province overlap among model-predicted candidates (OR≈2.34, p≈0.015); (5) fraction set robustness-the observed fractions rank at the 99.8th percentile among 1,000 random alternatives. A fine-grained scale scan identifies an empirical optimum at λ≈38,610 km (f≈7.76 Hz), reducing geometric mismatch by 17% compared to the theory-derived value; both fall within the Schumann resonance band (7.5-8.0 Hz). We adopt the working name Fraction-Network Interface-Zone Model (FNIZ) to emphasize that the geometry selects for macro-scale Earth-system interface landscapes (water-rock boundaries, basin margins, karst potential). The Mars-scaled version of the same geometry shows no enrichment for analogous interface features, consistent with an Earth-specific signal. All parameter choices were locked before validation tests; null controls and multiple-testing corrections are reported throughout.</p>