The LP procedure solves linear programs with a primal simplex solver. It can perform several types of post-optimality analysis, including range analysis, sensitivity analysis, and parametric programming. The procedure can also be used interactively.

PROC LP requires a problem data set that contains the model. In addition, a primal data set can be used for warm-starting a problem that has been partially solved previously.

The problem data that describe the model can be in one of two formats: dense or sparse. The dense format represents the model as a rectangular coefficient matrix. The sparse format, on the other hand, represents only the nonzero elements of a rectangular coefficient matrix.

For more details about the LP procedure, see Chapter 5: The LP Procedure.