int QSopt_primal (QSprob p, int *status)

In general, there is not a clear-cut winner between the primal and dual simplex algorithms. If performance is important in an application, it may pay to make tests of both methods to determine which function to use in the given setting.

After a call to QSopt_primal, the status variable should be checked to determine the status of the solution. In the most common cases, either an optimal solution is found (indicated by a value of QS_LP_OPTIMAL), or the problem is shown to be in infeasible (QS_LP_INFEASIBLE), or the problem is shown to be unbounded (QS_LP_UNBOUNDED). Optimization Status Codes gives a full list of the possible values for status.

The QSset_param function can be used to select the pricing rule that is used in QSopt_primal. The appropriate value for the whichparam argument of QSset_param is given below.

	Constant		Parameter
	QS_PARAM_PRIMAL_PRICING		Primal Pricing

The QSopt library has available four pricing rules for the primal simplex algorithm. The pricing rule can be specified in QSset_param by setting the newvalue argument to one of the values given below.

	Constant		Rule
	QS_PRICE_PDANTZIG		Primal Dantzig
	QS_PRICE_PDEVEX		Primal Devex
	QS_PRICE_PSTEEP		Primal steepest-edge

See Parameter Setting for the Optimizer for more parameters associated with the optimization algorithm

int status, rval;

/* p is a QSprob, a handle to an existing LP problem */

rval = QSopt_primal (p, &status);
if (rval) 
    fprintf (stderr, "QSopt_primal failed with return code %d\n", rval);
} else {
    switch (status) {
    case QS_LP_OPTIMAL:
        printf ("Found optimal solution to LP\n");
        break;
    case QS_LP_INFEASIBLE:
        printf ("No feasible solution exists for the LP\n");
        break;
    case QS_LP_UNBOUNDED:
        printf ("The LP objective is unbounded\n");
        break;
    default:
        printf ("LP could not be solved, status = %d\n", status);
        break;
    }
}