量子优化、量子模拟、光子集成、中微子 | 本周直播物理讲座

报告人：陈建鑫，阿里巴巴集团

时间：7月9日（周四）10:00-11:00

单位：中科院物理所

参加方式：讲座以线上形式进行，采用“腾讯会议”软件

会议ID：224 345 763
会议密码：0709
直播链接：https://meeting.tencent.com/l/YkZAQhaRtUlk

摘要：
It is believed that random quantum circuits are difficult to simulate classically. These have been used to demonstrate quantum supremacy: the execution of a task on a quantum computer that is infeasible for all classical computers. The task underlying the assertion of quantum supremacy by Arute {\em et al.} ({\em Nature}, {\bf 574}, 505--510(2019)) has been estimated to require Summit, the world's most powerful supercomputer today, at least 10,000 years. The same task was performed on their quantum processor Sycamore in only 200 seconds.
In this work, we present a classical simulation algorithm that, using a cluster on Alibaba Cloud with the same number of nodes and GPU specifications as Summit, can perform this task in 20 days. On smaller instances, we reduce the runtime from years to minutes, surpassing even Sycamore. The key ingredient is identifying and optimizing the ``stem'' of the computation: a sequence of pairwise tensor contractions that dominate the computational cost. This orders-of-magnitude reduction in classical simulation time, together with proposals for further significant improvements, indicate that achieving quantum supremacy may require a period of continuing quantum hardware improvements, without an unequivocal first demonstration.

报告人简介：

Dr. Jianxin Chen is a Quantum Scientist of Alibaba Group and Head of Quantum Computer Systems of Alibaba Quantum Laboratory (AQL), a division of Alibaba Group's global research institute DAMO Academy. Jianxin received his Bachelor's and Ph.D. degrees from Tsinghua University, both in computer science. Prior to joining Alibaba, he was a postdoctoral fellow jointly with Department of Mathematics and Statistics, University of Guelph and Institute for Quantum Computing at University of Waterloo and then a Hartree postdoctoral fellow in the Joint Center for Quantum Information and Computer Science at University of Maryland. Jianxin's research focuses on classical simulation of quantum systems and scalable compilation of quantum circuits.

报告直播二维码

www.koushare.com/video/byyy

报告人：李亚康

时间：7月7日（周二）15:00

单位：中科院高能物理所

会议室名称：青年科技工作者园地 

参会方式： 
1）客户端入会链接：http://caslink.ihep.ac.cn/meetingSchduler/invite/anno/0NHIFC5M
2）硬件终端拨打云会议室号 : 9005069149 

中国散裂中子源是研究中子特性、探测物质微观结构和运动的科研装置，基于该平台开展的中子散射研究可带动物理学、化学、生命科学、材料科学等多学科的发展。本次报告首先介绍中子散射实验的一般过程，然后介绍目前实验流程存在的一些问题，最后介绍深度学习在中子散射数据分析中的一些应用。 

报告人：付世园 

时间：7月7日（周二）15:00

单位：中科院高能物理所

会议室名称：青年科技工作者园地 

参会方式： 
1）客户端入会链接：http://caslink.ihep.ac.cn/meetingSchduler/invite/anno/0NHIFC5M
2）硬件终端拨打云会议室号 : 9005069149 

同步辐射装置产生的同步辐射图像是围绕样品中轴旋转获得的16bit投影图，具有高分辨率、高帧率的特点，HEPS预计每天将产生500TB数据，为存储造成极大压力。由于科学研究的需要，压缩过程不能丢失任何信息。但是，常见的图像无损压缩方法效果不佳。在此基础上，提出了一种基于图像差分和神经网络的图像无损压缩方法。采用图像差法减少图像与图像内部的线性相关性，然后使用神经网络结合差分结果的时间或空间上下文预测像素值分布，结合算术编码压缩。在分析压缩时间的基础上，将16位数据分成两部分进行并行压缩，以减少预测时间和编码时间。基于SSRF生成的图像的测试表明，该方法压缩结果优于PNG、JPEG2000和FILF，数据分割可以缩短预测和编码的时间。 

更多报告信息：中国物理学会期刊网学术讲座列表