Ratio Covers of Convex Sets and Optimal Mixture Density Estimation
Spencer Compton, Gábor Lugosi, Jaouad Mourtada, Jian Qian · 2026
Read Ratio Covers of Convex Sets and Optimal Mixture Density Estimation by Spencer Compton, Gábor Lugosi, Jaouad Mourtada, Jian Qian (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-2602-16142.txt.
Description
Spencer Compton leads a team of five math researchers who study density estimation under Kullback-Leibler divergence. Given samples from an unknown density, the goal is an estimate that stays close to the truth with high odds. Two settings matter. In model aggregation, the truth sits in a finite set of densities. In mixture estimation, the truth blends densities from that set. The authors assume nothing about the base densities. Ratios may be unbounded and supports may differ.
For both problems they pin down the best possible high-odds guarantees, which depend on set size, sample size, and confidence level. For mixtures they match known lower bounds for discrete cases. The mixture case hinges on two new covering results: a sharp bound on local Hellinger entropy for mixtures, and an optimal ratio covering theorem for convex sets. The latter also yields new estimates for approximate Pareto sets.
Catalog record
| CREATOR | Spencer Compton, Gábor Lugosi, Jaouad Mourtada, Jian Qian |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 89 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-16142