The Linear Programming Solver |
Pricing Strategies for the Primal Simplex Solver
Several pricing strategies for the simplex solvers are available. Pricing strategies determine which variable enters the basis at each simplex pivot. These can be controlled by specifying the PRICETYPE= option.
The primal simplex solver has the following five pricing strategies:
- PARTIAL
- scans a queue of decision variables to find an entering variable. You can optionally specify the QUEUESIZE= option to control the length of this queue.
- FULL
- uses Dantzig's most violated reduced cost rule (Dantzig 1963). It compares the reduced cost of all decision variables, and selects the variable with the most violated reduced cost as the entering variable.
- DEVEX
- implements the Devex pricing strategy developed by Harris (1973).
- STEEPESTEDGE
- uses the steepest-edge pricing strategy developed by Forrest and Goldfarb (1992).
- HYBRID
- uses a hybrid of the Devex and steepest-edge pricing strategies.
The dual simplex solver has only three pricing strategies available: FULL, DEVEX, and STEEPESTEDGE.
Copyright © 2008 by SAS Institute Inc., Cary, NC, USA. All rights reserved.