High Probability Complexity Bounds of Trust-Region Stochastic Sequential Quadratic Programming with Heavy-Tailed Noise
Yuchen Fang, Javad Lavaei, Sen Na · 2025
Read High Probability Complexity Bounds of Trust-Region Stochastic Sequential Quadratic Programming with Heavy-Tailed Noise by Yuchen Fang, Javad Lavaei, Sen Na (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-19091.txt.
Introduction
Yuchen Fang, Javad Lavaei, and Sen Na bound the work a stochastic optimizer needs when its noise has heavy tails. Their paper from twenty twenty-five studies problems with a random goal and fixed equality rules. Their method wraps sequential quadratic programming in a trust region. True values, slopes, and curves are not at hand; noisy oracles stand in.
Past studies assumed tame, light-tailed noise and first-order targets. This work allows biased, heavy-tailed noise in the value oracle. It also reaches second-order targets, points that pass a curvature test too. Rates match the light-tailed case for first-order points. Second-order points cost more steps, as theory expects. A floor on the target gap is set by the bias size. Tests on the CUTEst set check the method in practice.
Catalog record
| CREATOR | Yuchen Fang, Javad Lavaei, Sen Na |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 208 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-19091