한국경영과학회:학술대회논문집 (Proceedings of the Korean Operations and Management Science Society Conference)
- 한국경영과학회 2003년도 추계학술대회 및 정기총회
- /
- Pages.89-92
- /
- 2003
내부점 선형계획법에서의 사후처리
Postsolving in interior-point methods
초록
It is often that a large-scale linear programming(LP) problem may contain many constraints which are redundant or cause infeasibility on account of inefficient formulation or some errors in data input. Presolving or preprocessing is a series of operations which removes the underlying redundancy or detects infeasibility in the given LP problem. It is essential for the speedup of an LP system solving large-scale problems to implement presolving techniques. For the recovery of an optimal solution for the original problem from an optimal solution for the presolved problem, a special procedure, so called postsolving, must be applied. In this paper, we present how a postsolving procedure is constructed and implemented in LPABO, a interior-point based LP system. Briefly, all presolving processes are logged in a data structure in LPABO, and after the end of the solution method an optimal solution for the original problem is obtained by tracing the logs. In each stage of the postsolving procedure, the optimality of intermediate solutions is maintained. We tested our postsolving procedure on Netlib, Gondzio and Kennington LP data sets, and concluded that the computational burden of the procedure is relatively negligible compared with the total solving time.
키워드