报告人：中国人民大学 谢志远 教授

时间：2025-12-09（周二） 16:00-17:30

地点：科技楼A区311

摘要：Over the past two decades, tensor networks have emerged as an important tool in the study of statistical physics and condensed matter physics. Concurrently,neural networks have achieved great breakthroughs in computer vision and various machine learning tasks, and their potential for addressing complex physical problems has drawn growing attention from the physics community in recent years. Notably, the interplay between tensor networks and neural networks has evolved into a prominen research direction at this interdisciplinary frontier, fostering cross-fertilization between many-body physics and machine learning. In this talk, I will overview our group's recent exploratory studies on this interplay, including the development of differentiable tensor renormalization group methods, the construction of neural networks with matrix product operator and coarse-grained structures, and some preliminary results on eigenvalue problems in machine learning contexts.

报告人简介

谢志远，中国人民大学教授。2007年本科毕业于哈尔滨工业大学，2012年在中国科学院理论物理研究所获博士学位，随后在中科院物理所理论室从事博士后研究工作，2015年秋入职中国人民大学物理系，现任物理学院教授。报告人长期致力于张量重正化群方法的发展，及其在量子多体强关联系统、相变与临界系统中的应用，近年来关注张量网络与神经网络、量子计算、第一性原理计算理论等领域的交叉。

联系人 马天星 txma@bnu.edu.cn