Guardat en:
| Autor principal: | |
|---|---|
| Format: | Recurso digital |
| Idioma: | anglès |
| Publicat: |
Zenodo
2026
|
| Accés en línia: | https://doi.org/10.5281/zenodo.18439301 |
| Etiquetes: |
Afegir etiqueta
Sense etiquetes, Sigues el primer a etiquetar aquest registre!
|
Taula de continguts:
- <p>The Log-Rank conjecture posits that the complexity of a matrix is polynomially related<br>to the logarithm of its rank. We investigate this intuition through the lens of “Scholz<br>Arithmetic Cost”, an empirical metric derived from addition chain complexity. In density-<br>controlled Monte-Carlo simulations scaling up to N = 128, we report a robust decoupling<br>phenomenon: Dense low-rank matrices (r = 3) are information-theoretically trivial (LZMA<br>compression ratio ≈ 0.01), yet their normalized arithmetic construction cost remains statis-<br>tically indistinguishable from random noise ( eC ≈ 0.74). Regression analysis confirms that<br>the arithmetic cost is driven almost exclusively by bit-density (βdensity ≈ 0.51), with the<br>low-rank indicator having a negligible effect (βrank ≈ −0.0025). These findings suggest an<br>“Arithmetic Entropy Barrier”: a noise floor where algebraic simplification fails to reduce<br>arithmetic construction cost, serving as a critical caveat for complexity bounds based solely<br>on rank.</p>