An ABS approach for interior point methods for some special QCPs

摘要

We make use of the properties of the class of extended ABS (EABS) algorithms to present an efficient class of algorithms for computing the search directions of the primal{dual infeasible interior point methods (IIPMs) for solving convex quadratic programming problems (CQPs), when when the number of variables and constraints are equal. We show that, in this case, the parameters of the EABS algorithms for computing a search direction can always be chosen so that a part of the search vectors of the corresponding member of the class of EABS algorithms does not change in various iterations of the IIPMs. Keywords: Quadratic programming, Infeasible interior point methods, Primal-dual algorithms, ABS algorithms, Search direction.