Title :
A Primal-Dual Interior-Point Algorithm for Convex Quadratic Optimization Based on a Parametric Kernel Function
Author :
Wang, Guoqiang ; Bai, Yanqin
Author_Institution :
Coll. of Adv. Vocational Technol., Shanghai Univ. of Eng. Sci., Shanghai, China
Abstract :
In this paper a primal-dual interior-point algorithm for convex quadratic optimization based on a parametric kernel function, with parameters p isin [0,1] and q ges 1, is presented. The proposed parametric kernel function is of excellent properties which are used both for determining the search directions and measuring the distance between the given iterate and the central path for the algorithm. These properties enable us to derive the currently best known iteration bound for the algorithm with large-update method, namely, O(radicn log n log n/isin), which reduces the gap between the practical behavior of the algorithm and its theoretical performance results.
Keywords :
computational complexity; convex programming; iterative methods; quadratic programming; convex quadratic optimization; iteration bound; large-update method; parametric kernel function; primal-dual interior-point algorithm; Algorithm design and analysis; Convergence; Educational institutions; Kernel; Mathematics; Optimization methods; Polynomials; Prototypes; Symmetric matrices; Convex quadratic optimization; Interior-point methods; Iteration bound; Large-update method;
Conference_Titel :
Computational Sciences and Optimization, 2009. CSO 2009. International Joint Conference on
Conference_Location :
Sanya, Hainan
Print_ISBN :
978-0-7695-3605-7
DOI :
10.1109/CSO.2009.155