Abstract: In this talk, we present a class of numerical methods that can preserve simultaneously all invariant Hamiltonians of stochastic Poisson systems, based on discrete gradients and a projection technique. Structure-preserving properties of such methods are investigated and the root mean-square convergence orders of the schemes are analyzed. Numerical tests are performed to verify the theoretical results and illustrate the numerical behavior of the proposed methods.

Biography: Dr. Lijin Wang is currently an associate Professor at the School of Mathematical Sciences of University of Chinese Academy of Sciences. She received her Ph.D. in 2007 under joint education between the AMSS of the Chinese Academy of Sciences and the Karlsruhe Institute of Technology of Germany, and then she did postdoctoral research from 2007 to 2009 at the Tsinghua University of China. She has been working at the University of Chinese Academy of Sciences since 2009. Her research is mostly focused on numerical methods for stochastic differential equations, especially stochastic structure-preserving numerical methods for stochastic Hamiltonian systems, stochastic Poisson systems, etc..