arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2312.09686math.DGmath.CO
收藏

用于熵曲率的Bakry-Émery演算、新的直径估计与谱间隙

Bakry-Émery calculus for entropic curvature, new diameter estimates, and spectral gaps

Supanat Kamtue, Shiping Liu, Florentin Münch, Norbert Peyerimhoff

AI总结:

该研究推广Bakry-Émery演算以统一表述Bakry-Émery曲率与熵曲率,引入测度最优性准则与熵曲率的维度参数,并给出正熵曲率Markov链的直径估计及蕴含非负曲率扩展图不存在的谱间隙估计。

摘要:

本文提出了Bakry-Émery演算的一种推广形式,可同时构建Bakry-Émery曲率与熵曲率。该表述将两种曲率均表示为Bochner公式关于某测度的积分,由此引出测度的自然最优性准则,我们在Bakry-Émery框架下对其展开研究。此外,我们的方法还在熵曲率框架中引入了维度参数。我们还展示了梯度估计的应用,即具有严格正熵曲率的Markov链的直径估计,以及一项谱间隙估计,后者意味着不存在非负曲率的扩展图(expander)。

英文摘要:

In this paper, we propose a generalization of Bakry-Émery's calculus which allows us to formulate both Bakry-Émery and entropic curvature simultaneously. This formulation represents both curvatures as an integral of the Bochner formula against some measure. This leads to a natural optimality criterion of measures, which we investigate in the Bakry-Émery setting. Moreover, our approach leads also to a dimension parameter in the framework of entropic curvature. We also present gradient estimate applications, that is, diameter estimates for Markov chains with strictly positive entropic curvature and a spectral gap estimate. The latter implies non-existence of non-negatively curved expanders.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑