报告人：解家辉（博士，新加坡国立大学）

题目: Extreme eigenvalues of sample covariance matrices under generalized elliptical models with application in multiplier bootstrap.

摘要: We investigate the distributions of the first few largest eigenvalues of sample covariance matrices with separable structure $Q=TXD^2X*T*$. We show that depending on $D^2$, $TT*$ and the aspect ratio $p/n$, the edge eigenvalues can have various types of distributions asymptotically, including the three extreme value distributions for sequences of i.i.d. variables (cf. Gumbel, Frechet, Weibull), the TW law, Gaussian, or the mixture of TW law and Gaussian.

报告时间：2024年9月20日 16：00-17：00

报告地点：理学楼609

报告人简介:

解家辉，新加坡国立大学博士，现在新加坡国立大学从事博士后工作，研究兴趣包括Random matrices, high dimensional statistics等，已在概率统计学国际公认的顶尖杂志Biometrika，Bernoulli发表论文。