摘要: We investigate distributed methods for solving linear equations in multi-agent networks. Each agent only knows part of the overall equation and can communicate with its nearby neighbors. A distributed algorithm is devised to enable all agents to achieve exponentially fast a common solution. The proposed algorithm is distributed, works for all linear equations as long as the solution exists, does not involve any sufficiently small step size, and works asynchronously. Further improvement to the algorithm is also discussed including utilization the sparsity to reduce the state dimension, elimination of the initialization step, and generalization to achieving least square solutions. Applications of the algorithm includes large content delivery across vehicular networks, distributed  network localizations, and so on.​
 
报告人简介: Shaoshuai Mou has been working as an assistant professor at School of Aeronautics and Astronautics at Purdue University since August 2015. He received his bachelor and master degree in Harbin Institute of Technology in 2006 and 2008, respectively. He completed his Ph.D. study at Prof. A. Stephen Morse’s group in Electrical Engineering at Yale University in 2014. Then he worked as a post-doc at MIT for a year.  During his Ph. D. study, he held a position of visiting scholar at Australian National University and worked part-time for Yale Law School. He has received the Yale University Raymond John Wean Fellowship (2009) and the Chinese Government Award for Outstanding Students Abroad (2014). His research interests include distributed algorithms and control, multi-agent networks, formation control, and collaborations of UAVs.​