U.S.-Brazil Cooperative Research on Proximal Interior Point Methods

INT-9600343 Monteiro这个由佐治亚理工学院的Renato Monteiro和里约热内卢IMPA的巴西研究员Alfredo Iusem合作的美国-巴西合作研究项目将利用最近内点法来解决约束优化问题。要解决的问题包括超线性收敛，线性和凸二次规划中生成的对偶序列的收敛，原始序列的极限的表征，以及连续轨迹和PIP算法的近似版本的研究。这一发展将与经典内点方法的最新发展相平行。这项拟议的研究将在佐治亚理工学院和里约热内卢的数学家之间建立一种新的合作联系。拟议的努力将迄今独立处理的两个领域联系起来，应能更好地理解这一类重要问题，并提供新的、更有效的方法。***

INT-9600343 Monteiro This U.S.-Brazil collaborative research project between Renato Monteiro at Georgia Institute of Technology and the Brazilian researcher Alfredo Iusem of IMPA in Rio de Janeiro will utilize proximal interior point methods (PIP) for solving constrained optimization problems. Issues to be addressed include superlinear convergence, convergence of the generated dual sequence in linear and convex quadratic programming, characterization of the limit of the primal sequence, and study of continuous trajectories and approximate versions of PIP algorithms. The development will parallel recent developments in classical interior point methods. The proposed research will create a new collaborative link between mathematicians in Georgia Tech and Rio de Janeiro. The proposed effort, by bridging two areas so far treated independently, should provide better understanding and new and more effective methods for an important class of problems. ***