Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Primal barrier methods for linear programmming: Technical report

Technical Report ·
DOI:https://doi.org/10.2172/6005795· OSTI ID:6005795

The linear program min c/sup T/x subject to Ax = b, x greater than or equal to 0, is solved by the projected Newton barrier method. The method consists of solving a sequence of subproblems of the form min c/sup T/x /minus/ ..mu sigma..ln x; subject to Ax = b. Extensions for upper bounds, free and fixed variables are given. A linear modification is made to the logarithmic barrier function, which results in the solution being bounded in all cases. It also facilitates the provision of a good starting point. The solution of each subproblem involves repeatedly computing a search direction and taking a step along this direction. Ways to find an initial feasible solution, step sizes and convergence criteria are discussed. Like other interior-point method for linear programming, this method solves a system of the form AH/sup /minus/1/A/sup T/q = y, where H is diagonal. This system can be very ill-conditioned and special precautions must be taken for the Cholesky factorization. The matrix A is assumed to be large and sparse. Data structures and algorithms for the sparse factorization are explained. In particular, the consequences of relatively dense columns in A are investigated and a Schur-complement method is introduced to maintain the speed of the method in these cases. An implementation of the method was developed as part of the research. Results of extensive testing with medium to large problems are presented and the testing methodologies used are discussed. 51 refs.

Research Organization:
Stanford Univ., CA (USA). Dept. of Operations Research
DOE Contract Number:
FG03-87ER25030
OSTI ID:
6005795
Report Number(s):
SOL-89-6; ON: DE89014905
Country of Publication:
United States
Language:
English

Similar Records

Correlative Sparsity in Primal-Dual Interior-Point Methods for LP, SDP, and SOCP
Journal Article · Fri Aug 15 00:00:00 EDT 2008 · Applied Mathematics and Optimization · OSTI ID:21241987

A Schur-complement method for sparse quadratic programming
Technical Report · Thu Oct 01 00:00:00 EDT 1987 · OSTI ID:6011179

A Newton-type method for sparse quadratic programming problems
Conference · Fri Dec 30 23:00:00 EST 1994 · OSTI ID:36009