Skip to Content

物理前沿系列讲座之11:Manybody Localization, Quantum Thermalization, and Machine Learning

题目:Manybody Localization, Quantum Thermalization, and Machine Learning

 

主讲人:李潇 博士

 

研究方向:多体局域化和机器学习在凝聚态物理中的应用

 

时间:201843日(周二)下午3         

 

地点:物理机电大楼五楼咖啡厅

 

 

主讲人简介:李潇博士于2014年拿到美国德克萨斯大学奥斯汀分校(The University of Texas at Austin)的物理学博士学位,师从牛谦教授。目前是美国马里兰大学凝聚态理论研究中心的博士后研究员,主要从事二维材料中的拓扑和输运性质的研究。已发表多篇高水平学术论文,包括 NaturePhys. Rev. Lett Phys. Rev. B .

 

报告摘要Quantum localization has been a central subject of condensed matter physics. In the past decade, this subject has again become an area of intense research activities as people realized that (single-particle) Anderson localization may survive finite interactions, leading to a manybody localized phase. More importantly, when an isolated system is manybody localized, it strongly violates the familiar ergodicity hypothesis in quantum statistical mechanics, and fails to thermalize on its own. In this talk I will focus on a one-dimensional mutually incommensurate bichromatic lattice system which has been implemented in ultracold atoms to study quantum localization. We argue that without interactions, there exists a single-particle mobility edge (SPME) in the energy spectrum ], which is an energy that separates extended eigenstates from localized ones. Our theoretical work subsequently led to a first experimental observation of SPME in one-dimensional systems . We further study the properties of manybody localization in such a system when the interaction is turned on, and discuss its implications for a possible manybody mobility edge. In particular, we show that state-of-the-art machine learning techniques can help us understand the phase diagram in this intriguing manybody system.