kaiyun開云官方網(wǎng)站學術講座

海 報

講座時間： 2023年3月31日14:00-15:00

講座地點：騰訊會議號：732 744 926；密碼：0331

主講人簡介：

劉新為，1998年博士畢業(yè)于中國科學院計算數(shù)學與科學工程計算研究所?，F(xiàn)任河北工業(yè)大學理學院教授、學術委員會主任，數(shù)學和統(tǒng)計專業(yè)碩士生導師，控制和人工智能方向博士生導師。兼任中國運籌學會常務理事，中國運籌學會數(shù)學規(guī)劃分會常務理事，中國數(shù)學會計算數(shù)學分會理事，河北省運籌學會副理事長兼秘書長，APJOR、MMOR、PJO和《計算數(shù)學》等國際國內期刊編委。主要研究非凸非線性優(yōu)化算法及收斂性理論，在最優(yōu)化、計算數(shù)學和人工智能與控制等方向國際重要刊物發(fā)表多篇學術論文。主持完成和在研國家自然科學基金面上項目5項，參與國家基金重大研究計劃項目1項，是河北省政府特貼專家和國家自然科學獎會評專家。

講座內容簡介：

Title: A novel augmented Lagrangian for inequality constrained optimization and its application in linear programming

Abstract: We introduce a twice differentiable augmented Lagrangian for optimization with general inequality constraints. Our function is a combination of the augmented Lagrangian and the logarithmic-barrier technique, and is a generalization of the Hestenes-Powell augmented Lagrangian. The associated augmented Lagrangian method is proved to have strong global convergence, the capability of rapidly detecting the possible infeasibility, and linear convergence to the KKT point. The preliminary numerical experiments on some small benchmark test problems demonstrate our theoretical results. The application in linear programming shows its superiority.