Xiaozhou Li, Associate Professor, University of Electronic Science and Technology of China

Yong Liu, Associate Professor

Gaining Insight of Enhancing Accuracy with Convolution Filters

15:00-16:00 November 18(Monday), Z311

Filtering is a powerful tool in numerical simulations that can aid in accurately and efficiently predicting the governing physics in simulations, leading to improved designs. Its versatility is evident in various fields like data compression, shock filtering, post-processing, and machine learning. The increased adoption of filtering has emphasized the importance of designing effective convolution filters for diverse applications. This presentation delves into the design of convolution filters within the Smoothness-Increasing Accuracy-Conserving (SIAC) framework. A well-designed SIAC filter can uncover hidden information in certain numerical simulations, creating even more accurate representations of the data. These filters can be customized for boundaries, unstructured grids, and non-smooth solutions, offering adaptability to experimental data integration.

The main aim of this talk is to introduce the filtering technique, highlighting what it is, why it is useful. Furthermore, we demonstrate the usage of the filtering technique with existing code/software, identifying the essential properties in convolution filter design, what information it utilizes, and the potential applications it offers.

报告人简介:李小舟博士，2010年本科毕业于中国科学技术大学数学系，2015年于荷兰代尔夫特理工大学获得博士学位，2015年至2017在瑞士Università della Svizzera Italiana计算科学研究所、心脏病学计算中心任博士后。2017年9月至今在电子科技大学数学科学学院任副研究员。主要研究方向为微分方程的高阶精度数值方法（DG方法及其超收敛后处理技术等）、高性能计算等。