Lift zonoid order and functional inequalities
Alexei M. Kulik, Taras D. Tymoshkevych

TL;DR
This paper introduces weighted lift zonoids to establish new functional inequalities for measures, including extensions of Bobkov's shift inequality and a novel inverse log-Sobolev inequality, with potential applications to classical log-Sobolev inequalities.
Contribution
It defines weighted lift zonoids and demonstrates their use in deriving new functional inequalities, notably with a novel choice of weights for inverse log-Sobolev inequalities.
Findings
Derived non-linear extensions of Bobkov's shift inequality.
Introduced a new weighted inverse log-Sobolev inequality.
Suggested potential for proving classical log-Sobolev inequalities.
Abstract
We introduce the notion of a weighted lift zonoid and show that, for properly chosen weights v, the ordering condition on a measure \mu, formulated in terms of the weighted lift zonoids of this measure, leads to certain functional inequalities for this measure, such as non-linear extensions of Bobkov's shift inequality and weighted inverse log-Sobolev inequality. The choice of the weight K, involved in our version of the inverse log-Sobolev inequality, differs substantially from those available in the literature, and requires the weight v, involved into the definition of the weighted lift zonoid, to equal the divergence of the weight K w.r.t. initial measure \mu. We observe that such a choice may be useful for proving direct log-Sobolev inequality, as well, either in its weighted or classical forms.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
