Decomposition methods using compound proposals for large-scale optimization
- 98 Downloads
The approach presented here allows us to unite some advantages of the decomposition and basis factorization. First, we have a freedom to some extent to iterate in the master as in the decomposition methods. Second, the solution path goes similar to the one of the basis factorization. It also enables us to view the decomposition and basis factorization from a unifying position. The approach seems to be promising for parallel computations.
Unable to display preview. Download preview PDF.
- Grigoriadis, M.D. (1973) Unified pivoting procedures for large structured linear systems and programs. In Decomposition of Large-Scale Problems, D.M. Himmilblau, ed., North-Holland, 447–465.Google Scholar
- Kallio, M.J. (1975) On large-scale linear programming. Systems Optimization Laboratory, Stanford University, Technical Report SOL 75-7, Stanford, CA.Google Scholar
- Krivonozhko, V.E. (1991) On comparison of solution trajectories between Dantzig-Wolfe decomposition and basis factorization. Optimization Methods and Software, to appear.Google Scholar
- Murtagh, B.A. and Saunders, M.A. (1981) A projected lagrangian algorithm and its implementation for sparse nonlinear constraints. Technical Report Sol 80-1R, Stanford University, California.Google Scholar