Relaxed Triangle Inequality for Kullback-Leibler Divergence Between Multivariate Gaussian Distributions
Shiji Xiao, Yufeng Zhang, Chubo Liu, Yan Ding, Keqin Li Fellow IEEE, Kenli Li · 2026
Read Relaxed Triangle Inequality for Kullback-Leibler Divergence Between Multivariate Gaussian Distributions by Shiji Xiao, Yufeng Zhang, Chubo Liu, Yan Ding, Keqin Li Fellow IEEE, Kenli Li (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-2602-02577.txt.
Description
Shiji Xiao and 5 co-authors find the exact worst case for a relaxed triangle inequality between Gaussian distributions, in a 2026 arXiv paper. The Kullback-Leibler divergence measures how one probability distribution differs from another, but it is not a true distance and breaks the triangle inequality. Earlier work showed a relaxed version holds for multivariate Gaussians, with a loose constant.
Given two small divergence bounds along a chain of three Gaussians, the authors compute the supremum, the least upper bound, of the direct divergence. They also give the conditions under which that supremum can actually be attained. For small bounds the supremum tightens to a cleaner expression than the earlier estimate. The result is not just theoretical tidiness. The authors demonstrate applications in out-of-distribution detection with flow-based generative models and in safe reinforcement learning, where chained divergence bounds govern reliability.
Catalog record
| CREATOR | Shiji Xiao, Yufeng Zhang, Chubo Liu, Yan Ding, Keqin Li Fellow IEEE, Kenli Li |
|---|---|
| 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/2602-02577