Figure 4.54: The Preprocessing Tab in XPRESS Options Dialog Box
You can change the Preprocessing options for XPRESS by choosing XPRESS parameters from the Options menu and then pressing the Preprocessing tab. This will display the dialog box shown below:
Figure 4.54: The Preprocessing Tab in XPRESS Options Dialog Box
| Option Name | MPL Name | Solver Param | ParamNr | Type | Default | Min | Max |
|---|---|---|---|---|---|---|---|
| Use Presolve | Presolve | Presolve | 8011 | list | 1 | -1 | 2 |
| Trace Infeasibility | InfeasDiagCtlr | Trace | 8130 | int | 0 | -MAXINT | MAXINT |
| MIP Presolve | MipPresolve | MIPPresolve | 8078 | int | 0 | 0 | 7 |
The presolve facility can greatly improve performance by modifying the user's matrix so that it is easier to solve. The presolve algorithms identify and remove redundant rows and columns, thus reducing the size of the matrix. If the model contains global entities, integer presolve methods such as bound fixing and coefficient tightening are also applied to tighten the LP relaxation.
When a solution to a presolved problem has been found it is internally postsolved so that the user can access a solution to the original problem.
If Use presolve is On, the optimizer will begin by preprocessing the matrix. This reduces the active matrix size by identifying redundant rows and columns by inspection and modifying the matrix in order to make it easier to optimize.
| Return infeasible (-1) | Presolve applied, but a problem will not be declared infeasible if primal infeasibilities are detected. The problem will be solved by LP optimization algorithm, returning an infeasible solution, which can sometimes be helpful. |
| Presolve not applied (0) | Presolve not applied. |
| Presolve applied (1) | Presolve applied. |
| Redundant bounds not removed (2) | Presolve applied, but redundant bounds are not removed. This can sometimes increase the efficiency of the barrier algorithm. |
Control of the infeasibility diagnosis during presolve. If On, the logical deductions made by presolve to deduce infeasibility will be displayed.
If any of these options are selected, integer preprocessing for MIP problems will be performed at each node of the branch-and-bound search tree (including the top node). If a variable is fixed at a node, it remains fixed at all its child nodes but it is not deleted from the matrix (unlike the variables fixed by presolve).
| Reduced Cost Fixing (1) | Reduced cost fixing at each node. |
| Logical preprocessing (2) | Logical preprocessing at each node. |
| Probing (at root only) (4) | Probing set at the top node. |