Speaker: 卢丽强，浙江大学

Inviter: 尚云研究员

Title: 从张量计算到量子计算的一些思考

Language: Chinese

Time & Venue: 2024.12.11 09:30-10:30 南楼N224

Abstract: Tensor computing and quantum computing are two representative diagrams of advanced computing. Specifically, tensor computing serves as the most important operator in recent AI applications, while quantum computing has the potential to revolutionize the computational infrastructure. This talk summarizes the progress of Dr. Lu in these two fields, starting from his Ph.D. career to recent progress at Zhejiang University (ZJU). First, I will introduce an analytical modeling framework that helps to guide the architecture design for tensor-based applications. This includes a relation-centric dataflow representation, which formally describes the relationship between computational instances and hardware behavior. I will then present several chip designs for popular AI applications. Next, I will discuss how I move to quantum computing and present the progress made in recent two years. This includes efforts in full-stack optimization for quantum computer architecture, focusing on error mitigation, compilation optimization, quantum program verification, and software-hardware co-design.way for efficiently realizing quantum simulations and demonstrating advantages on near-term quantum devices.