报告人： 韩放 讲席教授

报告题目： Limit Theorems of Azadkia-Chatterjee’s Conditional Graph Correlation

时间： 2026年7月27日（周一）10:00—12:00

地点： 西安交通大学兴庆校区数学楼2-3会议室

邀请人： 姜丹丹 教授

报告摘要：
Azadkia and Chatterjee’s ingenious rank- and graph-based approach to measuring conditional dependence strength has inspired a large and rapidly growing body of work. Despite this progress, the fundamental statistical properties of the resulting graph correlation remain largely unresolved. In this talk, I will present a complete set of results describing its asymptotic behavior, including a CLT, a principled bias-correction procedure, and a near-linear-time estimator of its limiting variance. Taken together, these results make statistical inference based on Azadkia-Chatterjee’s conditional graph correlation, including confidence interval construction and conditional independence testing, possible for the first time.

报告人简介：
韩放，现任美国华盛顿大学（University of Washington, Seattle）统计系 Job & Gertrude Tamaki 讲席教授。他于2008年获北京大学数学学院博士学位。韩放教授的主要研究方向为理论统计与计量经济学，现任 Annals of Statistics 与 Bernoulli 副主编。