腾讯会议：7568896432
邀请人：姜波副教授
报告摘要：We propose an adaptive trust-region truncated three-term conjugate gradient method for Riemannian optimization problems. Especially, the trust region radius converges to zero with the adaptive technique， and the trust region subproblem is solved by using the truncated three-term conjugate gradient method with a new restart strategy. We present some properties of the present Riemannian method and establish the global convergence and local superlinear convergence under the Lipschitz condition with regard to isometric vector transport. Comparing numerical results obtained for the Rayleigh quotient minimization problem, Principal Component Analysis problem, joint diagonalization problem are reported to demonstrate the e_ectiveness of the proposed Riemannian method.
报告人简介：
严涛，南京理工大学数学与统计*女王调教-女王调教视频-女王 调教小说副教授、硕士生导师。主要研究方向为流形优化算法及其应用、均衡约束规划。在中国科学、Computerand Mathematics with Application、Optimization等杂志发表论文多篇。完成国家自然科学基金面上项目1项，国家自然科学基金青年项目1项。