Metraipolis › archive › dc_e1e3552b3ceb

The Precision Wall: An Optimizer-Dependent Boundary for Optimization under Power-Law Measurement Spectra — arXiv:2610.40485

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.40485
License (as archived)
CC0-1.0
Archived
2026-11-01T02:26:47.137Z
SHA-256 of the archived text
bc5deb84d45614aaece20c51eed5d207fa44954f113f2c1af202f5c3b82bb08d
Stable URL of this page
https://metraipolis.com/documento/dc_e1e3552b3ceb

How to cite

The Precision Wall: An Optimizer-Dependent Boundary for Optimization under Power-Law Measurement Spectra — arXiv:2610.40485 — arXiv (en), https://arxiv.org/abs/2610.40485 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e1e3552b3ceb (Metraipolis archive dc_e1e3552b3ceb, sha256 bc5deb84d45614aaece20c51eed5d207fa44954f113f2c1af202f5c3b82bb08d)
BibTeX
@misc{dce1e3552b3ceb,
  title = {The Precision Wall: An Optimizer-Dependent Boundary for Optimization under Power-Law Measurement Spectra — arXiv:2610.40485},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.40485},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e1e3552b3ceb; sha256 bc5deb84d45614aaece20c51eed5d207fa44954f113f2c1af202f5c3b82bb08d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40485. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

In low-rank matrix sensing of a rank-$D$ factor $U^\ast \in \mathbb{R}^{V \times D}$, replacing isotropic Gaussian designs with power-law measurement covariance $\Sigma = \mathrm{diag}(j^{-s})_{j=1}^V$ leaves the Burer--Monteiro landscape free of spurious local minima, yet division-normalised first-order optimizers stall above a critical exponent $s^\ast$. We trace this failure to a numerical precision wall. At the ground truth, the smallest non-gauge expected Hessian eigenvalue scales at leading order as $V^{-(s+1)}D^{-s}$. When this curvature drops below the effective denominator floor $\epsilon_O$ of optimizer $O$, normalisation noise overwhelms rare-direction gradients, yielding the leading-order boundary $s^\ast_O(V, D) = [\log(1/\epsilon_O) - \log V]/\log(VD) + \delta_O$, where $\delta_O$ is a metric-dependent detection lag. With $\delta_{\mathrm{Adam}}$ calibrated on four configurations and frozen, this formula predicts Adam's onset within $0.45$ across all 16 cells of a held-out grid ($V \in \{64,128,256,512\} \times D \in \{4,8,16,32\}$), separates fixed-$VD$ pairs, and distinguishes denominator-precision walls from conditioning and sign-quantisation walls. The approximation fails at $D = 2$, reflects a finite-$D$ spectral crossover, and saturates below $\varepsilon_0 = 10^{-8}$ at a dynamical floor.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.40485 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-01T02:12:44.218Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).