走向现代数学学术报告 - 肖传福助理研究员(No. 779)
报告题目:Tensor-based methods for solving high-dimensional problems in scientific computing
报告人:肖传福 助理研究员(北京大学长沙计算与数字经济研究院)
报告时间:2025年4月18日 9:00
报告地点:东海岸校区-D实209
报告摘要:There are many high-dimensional problems in the field of scientific computing, e.g., high physical dimension, high frequency, and inverse problems. Tensor as an efficient dimension reduction tool, has received widespread attention in various fields, such as computational physics/chemistry, signal processing, and machine learning. In this talk, I will introduce how to use tensor representations to address the challenge of curse of dimensionality. Moreover, I will use two examples, Boltzmann-BGK and matrix inversion to illustrate the advantages of tensor-based methods.
个人简介:肖传福,北京大学长沙计算与数字经济研究院助理研究员。2022年6月博士毕业于北京大学数学科学学院。同年7月,加入北京大学大数据国家工程实验室从事博士后研究工作,2024年7月加入北京大学长沙计算与数字经济研究院。主要研究兴趣包括矩阵/张量计算及其应用、模型约简、并行计算,致力于发展“张量表示+机器学习”工具并将其应用于解决不同领域中的实际问题,尤其是高维带来的挑战;相关研究成果主要发表在SIAM Journal on Matrix Analysis and Applications (SIMAX), Journal of Scientific Computing (JSC), International Conference on Learning Representations (ICLR)等国际计算数学期刊和人工智能会议上。