*讲座内容简介：
Many practical applications involve a rich spectrum of networks, which are modeled by graphs. In many of these applications, such as social, biological, or energy networks, the solution of large-scale graph Laplacians is a key ingredient of the computational model. Therefore, developing effective algorithms for such systems is an important and challenging task. In this talk, several recent advancements of the adaptive algebraic multigrid method for solving graph Laplacians are introduced, including building the multilevel structure via path cover and computing a posteriori error estimator based on Helmholtz decomposition on Graphs. Numerical results for solving real-world network applications are presented to demonstrate the efficiency of the proposed adaptive algebraic multigrid method.

*主讲人简介：
胡潇喆，美国塔夫茨大学数学系副教授，浙江大学本硕博，2009年获计算数学博士学位后在北京大学做博士后研究，2010年后在宾夕法尼亚州立大学数学系做研究，随后到塔夫茨大学工作。胡教授主要研究数值分析与科学计算, 特别是分析和设计数值算法求解偏微分方程及图论问题（背景涉及多学科：多孔介质流、生物信息学及机器学习等），已发表学术论文50多篇。此外，他的成果已被英伟达、中国海洋石油总公司和中国石油等企业所接纳和应用。2016年，他的成果在国际分数阶微积分及其应用大会上获得黎曼-刘维尔奖。其研究成果也是Disease Module Identification DREAM Challenges中表现最好的算法之一。同时他的研究得到美国能源部和美国国家科学基金会的资助。此外，也担任期刊《Numerical Linear Algebra with Application》和《Petroleum Science》的编委。