The smallest singular value for rectangular random matrices with Lévy entries
Electronic Journal of Probability · 01.01.2026 · 31 (none) · 1-29
DOI: 10.1214/26-ejp1628 ↗Kurz gefasst
Fachartikel in der Zeitschrift Electronic Journal of Probability, erschienen am 01.01.2026. Verfasst von Yi Han. Die Arbeit nennt 40 Literaturangaben.
Abstract (Originalsprache)
Let X=(xij)∈RN×n be a rectangular random matrix with i.i.d. entries (we assume N∕n→a>1), and denote by σmin(X) its smallest singular value. When entries have mean zero and unit second moment, the celebrated work of Bai-Yin and Tikhomirov show that n−1 2σmin(X) converges almost surely to a−1. However, little is known when the second moment is infinite. In this work we consider symmetric entry distributions satisfying P(|xij|>t)≍t−α for some α∈(0,2), and prove that σmin(X) can be determined up to a log factor with high probability: for any D>0, with probability at least 1−n−D we have C1n1 α(logn) 2(α−2) α≤σmin(X)≤C2n1 α(logn)α−2 2α for some constants C1,C2>0. The upper bound was derived in a recent work of Bao, Lee and Xu [7] but the lower bound is new and answers a problem posed in that paper in a weaker form. This appears to be the first determination of σmin(X) in the α-stable case with a correct leading order of n, as previous anti-concentration arguments only yield lower bound n1 2. The same lower bound holds for σmin(X+B) for any fixed rectangular matrix B with no assumption on its operator norm. The case of diverging aspect ratio is also computed.
Zitieren
APA
Han, Y. (2026). The smallest singular value for rectangular random matrices with Lévy entries. Electronic Journal of Probability, 31(none), 1-29. https://doi.org/10.1214/26-ejp1628
Vancouver
Han Y. The smallest singular value for rectangular random matrices with Lévy entries. Electronic Journal of Probability. 2026;31(none):1-29. doi:10.1214/26-ejp1628
BibTeX
@article{Han2026,
author = {Han, Yi},
title = {The smallest singular value for rectangular random matrices with Lévy entries},
journal = {Electronic Journal of Probability},
year = {2026},
volume = {31},
number = {none},
pages = {1--29},
doi = {10.1214/26-ejp1628}
}
EUVIMED