报告主题:求解正交约束最优化问题的并行要领
报告人:刘歆 副研究员 (中国科学院数学与系统科学研究院)
报告时间:2017年6月3日(周六)14:30
报告所在:校本部GJ408
约请人:余长君
主理部分:理学院数学系
报告摘要:To construct parallel approach for solving orthogonal constrained optimization problems is usually regarded as impossible mission, due to the low scalability of orthogonalization procedure. In this talk, we propose a Jacobi type column-wise block coordinate descent method for solving a class of orthogonal constrained optimization problems, and establish the global iterate convergence to stationary point of our proposed approach. Distributed algorithms are consequently implemented. Numerical experiments illustrate that the new algorithms have brilliant performance and high scalability in solving discretized Kohn-Sham total energy minimization problems.
接待西席、学生加入 !