betaFeedback

WorksMathematics

Generalized and Convex Poincaré Inequalities via Transport-Divergence Inequalities

Abstract

We investigate the relationship between generalized (q′,p′)(q',p')-Poincaré inequalities and transport-divergence inequalities involving χqχ^q-type divergences. Extending the classical correspondence between the (2,2)(2,2)-Poincaré inequality and quadratic transport-χ2χ^2 inequalities, we prove that a (q′,p′)(q',p')-Poincaré inequality is equivalent to a transport-divergence inequality for conjugates (p,p′)(p,p') and (q,q′)(q,q'). We then study the convex counterpart of these inequalities. Using weak barycentric optimal transport and convex order, we establish an equivalence between convex (q′,p′)(q',p')-Poincaré inequalities and weak transport-divergence inequalities involving a projected χqχ_q divergence. We further characterize the minimizers arising in this projection problem through a geometric condition on the relative density. In dimension one, we analyze the geometric consequences of (q′,p′)(q',p')-Poincaré inequalities by identifying sharp conditions on the behavior of the density both at infinity and near isolated zeros. We show that when q′>p′q'>p', the concentration behavior becomes polynomial rather than exponential at infinity. Finally, we interpret these inequalities through the lens of optimization in Wasserstein space. In particular, the equivalence between transport-divergence and Poincaré inequalities appears as an analogue of the equivalence between quadratic growth and Polyak–Łojasiewicz inequalities in finite-dimensional optimization.

Paper details

vlnc:W_2gu2jjc2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC-BY-NC-SA-4.0
Usage

views · 3

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

AI

View statement
Looking for related papers…