Bregman Meets Lévy: Stochastic Mirror Descent with Heavy-Tailed Noise in Continuous and Discrete Time
Pierre-Louis Cauvin c,∗, Panayotis Mertikopoulos ∗ · 2026
Read Bregman Meets Lévy: Stochastic Mirror Descent with Heavy-Tailed Noise in Continuous and Discrete Time by Pierre-Louis Cauvin c,∗, Panayotis Mertikopoulos ∗ (2026) 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-2606-03769.txt.
Description
Cauvin and Mertikopoulos study stochastic mirror descent under heavy-tailed noise. Gradient noise with infinite variance can break standard convergence proofs. The question is whether mirror descent keeps its guarantees in that regime. A continuous-time model frames the answer. Mirror descent becomes a stochastic differential equation driven by Lévy noise. Finite moments of order p, between one and two, are assumed.
The scaling limit is named the Lévy mirror flow. Its paths can jump by any size when p is below two. Despite that wild behavior, near-optimality arrives in a set time bound. The convex case and the strongly convex case each get explicit rates. Matching discrete-time guarantees follow for several mirror descent variants. The results show exactly how frequent long jumps slow convergence. Optimization readers get a robust map of a noisy frontier.
Catalog record
| CREATOR | Pierre-Louis Cauvin c,∗, Panayotis Mertikopoulos ∗ |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 203 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.03769