Deep learning-based method for solve incompressible Navier-Stokes equation and Cahn-Hillard equation


Speaker:    Qiaolin He, Sichuan University

Time:         10:30-11:30 , Jun .29

Location:   Tencent Meeting

    Meeting ID810-472-566


Host:          Yue Qiu



The speaker extends the algorithm presented by Han et al. to Navier-Stokes and Cahn-Hilliard equations in high dimension, which is an initial boundary value problem. The equation is reformulated using backward stochastic differential equations and the gradient of the unknown solution is approximated by neural networks. Numerical examples show the accuracy of the algorithm, which is quite effective in high dimension.



Dr. He obtained her bachelor and PhD degrees both in mathematics from Sichuan University and Hongkong University of Science and Technology (HKUST) in 2000 and 2007, respectively. She did her postdoc research at HKUST from 2007 to 2010. Since 2010 she has been working at School of Mathematical Sciences of Sichuan Universityand currently as professor. Her research interest focuses on computational fluid dynamics, scientific machine learning, and her research has been supported by NSFC and National Key Research and Development Program of China. Her research output has been published at ZAMM, DCDS. B, JCP, JCAM et al.